In the Matter of Nord Anglia Education Inc - Judgment

¶ IN THE GRAND COURT OF THE CAYMAN ISLANDS FINANCIAL SERVICES DIVISION CAUSE NO. FSD 235 OF 2017 (IKJ) IN THE MATTER OF SECTION 238 OF THE COMPANIES LAW (2016 REVISION) AND IN THE MATTER OF NORD ANGLIA EDUCATION, INC IN COURT Appearances: Lord Grabiner QC and Mr Richard Boulton QC of counsel and Mr Mac Imrie, Mr Malachi Sweetman and Mr Lukas Schroeter of Maples and Calder on behalf of Nord Anglia Education, Inc. ("the Company") Mr Richard Millett QC of counsel and Mr Simon Dickson, Ms Jessica Vickers and Mr Harry Rasmussen of Mourant Ozannes on behalf of the Mourant Dissenting Shareholders Mr Jonathan Adkin QC of counsel and Mr Andrew S Jackson of Appleby on behalf of the Appleby Dissenting Shareholders Mr George Bompas QC of counsel and Mr Hamid Khanbhai of Campbells on behalf of the Campbells Dissenting Shareholders (together "the Dissenters") Before: The Hon. Justice Kawaley Heard: 2 December – 20 December 2019 Draft Judgment Circulated: 28 February 2020 Judgment Delivered: 17 March 2020 HEADNOTE Appraisal of fair value of shares-section 238 of the Companies Law Petition-appropriate valuation method-whether market value or discounted cash flow analysis is appropriate valuation methodology-efficiency of market in the Company’s shares-relevance of transaction price-rigour of transaction process-whether merger agreement an arms-length transaction-relevance of non-public information-minority discount JUDGMENT INTRODUCTORY

¶ [1] On November 9, 2017, the Company presented a Petition under section 238(9) of the Companies Law (2016 Revision) for the determination of the fair value of the Dissenters’ Shares. The Shares had been acquired by the Company pursuant to a merger agreement dated April 25, 2017 ("Merger Agreement") which was approved by the requisite majority of Shareholders at an extraordinary general meeting held on August 21, 2017 ("EGM"). The path to trial has been strewn with many (often flowery) interlocutory applications, the history of which need not be recited here. In the final event the parties cooperated very effectively to ensure that the scheduled trial was completed on time, despite the multiplicity of witnesses who appeared and the range of issues which were canvassed by counsel.

¶ [2] The Company initially offered the Dissenters the Merger Consideration of US$32.50 for their Shares, before the present proceedings were commenced. At trial the Company contended that a market price of US$30.45 reflected the fair value of the Shares; the Dissenters contended that US$76.51 was, applying a discounted cash flow ("DCF") valuation methodology, the fair value of the Shares. THE ISSUES

¶ [3] There was no material dispute on the general legal principles governing the Court’s jurisdiction to determine the fair value of the Shares under section 238 of the Companies Law. How those principles should be applied to the facts of the present case, which were disputed to a material extent, in light of the expert evidence formed the central controversy.

¶ [4] The governing legal approach was summarised in the Company’s Written Opening Submissions as follows: “11. The meaning of fair value in the Cayman Islands has not been clarified by a precise legal definition. Nevertheless, the main concepts are becoming easier to express with the benefit of recent authority. For introductory purposes, the following points are the key ones: 11.1 It is the Dissenters' shares in the Company that are being valued; 11.2 The valuation is intended to compensate the Dissenters for the fair value of their shares and reflects an economic exchange of the rights and obligations attaching to the shares for cash; 11.3 'Fair' adds the concepts of just and equitable treatment, and flexibility, to 'value'; 11.4 Fair value applies to both the Dissenters and the Company; neither side should be preferred, and fair value does not require that the price must be the highest possible; 11.5 Excluded from fair value are the benefits and burdens of the merger transaction itself; 11.6 Minority shareholdings are to be valued as such, subject to the particular rights and liabilities attaching to the shares. 12 Consistent with these principles, in this case the experts have proceeded on the basis that fair value is to be expressed on a "per share" basis and have agreed that fair value refers to the value of a share in the Company as a going concern, and does not include advantages which accrue to the Company as a result of the merger, including anticipated synergies, and that minority shareholdings are to be valued as such.”

¶ [5] The Written Opening Submissions for Trial on behalf of the Dissenters described the governing legal principles concisely as follows: “2. The term “fair value” has a well-established meaning in Cayman law. As will be set out in detail below, it does not mean “merger price” or “market value” or even “fair market value”. Rather, it means what has been described as the intrinsic value of shares in the Company, valued as a going concern and taking into account all value-relevant information, whether or not available to the market, whilst leaving out of account any factors peculiar to a particular buyer or seller but irrelevant to the value of the business, and any enhancement or reduction in value arising from the merger itself. The statutory meaning of "fair value" is the only matter of law in this case, and the law is well established. The rest is the application of the evidence."

¶ [6] The main headline factual issues in controversy were the following: (a) whether the adjusted Market Price was the most appropriate valuation. The Company’s primary position was that US$30.45 reflected the fair value of the Shares. It was common ground that it would only be appropriate to use this if: (i) the market was semi-strong efficient; and (ii) there was no material non-public information ("MNPI") about the value of the Shares; (b) what, either as a check on the validity of the Market Price or because the Market Price was not an appropriate methodology to use, was the fair value of the Shares using the DCF valuation method. The Dissenters’ primary position was that their Expert’s DCF valuation should be adopted by the Court. Both Experts carried out what each contended was the correct valuation exercise but generated starkly different results: (i) $33.23 per share (Professor Fischel’s mid-range figure); (ii) $76.51 per share (Professor Gompers); (c) the critical elements for an appropriate DCF analysis were agreed in principle but the precise values to be applied to each input on the facts of the present case were contentious. The key elements are: (i) central or reasonable estimates of projected cash flows; (ii) a discount rate that reflects the risks attached to those projections, based on the Company’s weighted average cost of capital ("WACC"); (iii) a perpetuity growth rate based on central or reasonable estimates of long-term growth; and (d) whether the Merger Consideration or Take–Private transaction price of $32.50 (the "Transaction Price") was a relevant indicator as to the fair value of the Shares. The Company contended that the Transaction Price included a premium but was a useful indicator of fair value while the Dissenters contended that the process adopted fell short of being an arms’ length transaction. THE COMPANY’S FACT WITNESSES Graeme Halder Overview of witness

¶ [7] Mr Graeme Halder was at all material times Chief Financial Officer of the Company. A Chartered Accountant, he was responsible for the preparation of the Company’s financial statements and the financial projections which were central to the contentious DCF valuation analysis. He left the Company in August 2018.

¶ [8] I found Mr Halder to be a credible witness who generally gave his evidence in a careful and (in the face of probing cross-examination) impressively objective manner. His capacity for answering questions with precision is illustrated by the following exchanges during cross-examination by Mr Adkin QC: “...So it looks like early part of December the Project Darwin process is halted; yes? A. I mean, obviously I'm not copied on any of the emails, so I can't comment on those, but I was aware that, you know, the Project Darwin did come to a halt in either late November or early December. Q. I see. A. So, yes, I'm aware that that happened but not through, obviously, this process. I wasn't copied in on any of these emails. Q. I see. And were you involved in the decision to halt it? A. No. Q. Do you know who was involved in that decision? A. No. Q. Do you know why that decision was taken? A. Well, I can't if I did not know who was involved in taking it. Q. I think it's probably logically possible, but anyway, your answer is no, you don't? A. Correct."1

¶ [9] His ability to respond to potentially contentious questions in a balanced, fair yet firm manner is provided by the following extract from the Transcript: “A. That is what EY are saying, yes. Q. That's what they are saying. And if we go to page 4, {HSD/18/4}, there is a number of quick wins, so we have got food and catering, books, IT, it seems to be around there, facilities and management on the edge. Yes, do you see that? A. Yes, I can see it. Q. And this was pretty realistic, wasn't it? There were some pretty -- there is some low hanging fruit, where savings could be made? A. I mean, you know, I think there was an opportunity in procurement to make savings. I think there weren't quick wins, as we discovered when we came to actually implement it, which was later on in this year. You know, to change the way a school operates -- it's a fairly individual place, it operates as it wants to operate per the principal, which is one of the challenges of managing a global education business and trying to get them to buy into the idea of, you know, that procurement is a foreign language that you have to work hard to try and get them to understand. So, you know, my view is I think there was an opportunity. I think there is an opportunity. I'm not sure how much of it has been got as of today but the idea of quick wins I thought was always going to be difficult because, you know, there is absolutely no doubt about the analysis that says there are opportunities.”2

¶ [10] Perhaps the most important contentious aspect of his oral evidence was Mr Halder’s characterisation of the approach he followed in preparing forecasts of future performance: 1 Transcript Day 2 page 34 line 6-page 35 line 1. 2 Transcript Day 2 page 169 line 17-page 170 line 22. "Q...And what it says is that you had told Houlihan Lokey that the projections you gave them had been reasonably prepared -- let's just ignore the word "reasonable", if you like -- that they had been prepared in good faith on bases reflecting the best currently available estimates and judgments of management as to the future financial results and condition of the company, and that is indeed the basis upon which they had been prepared, isn't it? A. Yes, the information at that point in time, yes. Q. Yes. What you had done was produce projections that reflected what you believed in good faith, quite honestly, to be the best currently available estimates and judgment -- your estimates and your judgment -- as to the future financial results and condition of the company; yes? A. Yes, based on the way I produce forecasts. Q. Well, what do you mean by that? A. I mean, when I'm producing -- you know, so what I would like to say is, you know, when I was producing these forecasts, the one thing I didn't envisage would be sitting here two and a half years later discussing the forecasts. I approached forecasts very similarly throughout my career for 30 years, that you mentioned, that I did forecasts and I defined it in the responses to some of the letters as ambitious but achievable, and I've also talked about the fact that I did not have risk weighting within that, and that's the way that I did forecasts, you know. They were always stretching. That's what a CFO does in a business, he does stretching forecasts, stretching targets to push the business to try and achieve. And I -- you know, these forecasts were no different to the August 16 forecasts I did or the November 16 forecasts I did or the April -- sorry, the August 17 projections I did or the ones I did for the IPO or any other projections I did at Nord Anglia or anywhere else. They were stretching, they were ambitious but, in my view, achievable. So I think that's sits within that wording that I had created forecasts that were my best view of, you know, what the company could achieve at that point in time. What I didn't do was put in a risk weighting. And I've discussed why I didn't do that in my affidavit; it's difficult -- it's difficult to know how to risk-weight things that you don't know, that invariably are probably going to happen in the future but haven't yet happened. And I can give, you know, a myriad of examples of that, things that have happened in the past that, when we did previous five year projections we didn't know about, you know, some of which are pretty material."3

¶ [11] Save as regards the narrow issue of this portion of his evidence relating to the basis on which he prepared his projections, his credibility was not directly challenged. Even the sole challenge was advanced in a very nuanced way. I shall return to this issue in further detail below.

¶ [12] Overall Mr Halder appeared to me to give his evidence in a level-headed non-partisan manner. He briefly departed from this approach after lunch on the day of his testimony, causing me to speculate whether during the short adjournment he had been subjected to a passing jibe about giving his evidence in an overly 'passive' manner. Under cross-examination by Mr Adkin QC about future earnings potentials at around 3.00pm, he responded as follows: "...So that's a pretty promising outlook for the China Bilingual market, is it not, being expressed by Parthenon in this March 17 report? A. I mean, it might be that Ian Johnson was involved in either the agenda or reviewing this. So he might be a better person to ask because he was -- Q. I will do. A. -- directly responsible for that. You know, what I would say about it is I'm sitting here thinking I wonder why we paid them the money because it's pretty useless because if you look on the previous page, if we can just go back, and I read the cities that they reviewed and I don't know why -- sorry, the one before -- why they chose the cities they chose, and Ian will be able to correct me..."4

¶ [13] A few minutes later, the witness was forced to admit that the Company itself had requested Parthenon to study the cities they chose to study5. This was a very minor error 3 Transcript Day 2 page 49 line 19-page 51 line 17. 4 Transcript Day 2 page 153 line 13-page 154 line 2. 5 Transcript Day 2 page 155 line 8- page 156 line 7. on a matter as to which the witness admitted he had limited information, but it struck me at the time as flowing from the witness atypically providing a loose answer designed to bolster the Company’s case. Thereafter, it appeared to me, Mr Halder resumed the predominantly careful and neutral manner in which he gave his oral evidence. Overview of evidence-in-chief

¶ [14] Mr Halder’s First Affidavit was sworn on January 31, 2018 ("First Halder Affidavit") when he was still resident in Hong Kong and employed as Chief Financial Officer of the Company. For present purposes, this Affidavit pertinently supported the Petition by explaining the mostly non-contentious background to the Merger Agreement and the Company’s business. In summary: (a) the Company’s main business involved managing international schools from the pre-school to secondary level. At that time, the Company operated 54 schools in 24 countries with a student enrolment of over 49,000 and staff of over 9,000; (b) the Company’s business originated (under a different corporate structure) in 1972 from a base in the United Kingdom. The Company’s centre of management has been located in Hong Kong since 2012. The Company’s initial public offering on the New York Stock Exchange ("NYSE") took place on March 26, 2014 and the Shares were listed until September 1, 2017; (c) since 2008 the Company’s majority shareholder has been Premier Education Holdings ("Premier"). Some 96% of Premier’s shares were held by The Baring Private Equity Fund III and the Baring Private Equity Fund IV (the "Selling Funds"/"Fund III" and "Fund IV"). The Selling Funds are affiliated with “one of the largest independent alternative investment management firms in Asia”6, Baring Private Equity Group Asia Limited ("BPEA"); (d) the Merger Transaction was an arms’ length take-private acquisition transaction between the Company and a buyer consortium comprising: (i) The Baring Asia Private Equity Fund VI, L.P. 1, The Baring Asia Private Equity Fund VI, L.P. 2 and The Baring Asia Private Equity Fund VI Co-investment L.P. (the "Buying Funds"/"Fund VI LP 1", "Fund VI LP 2" and "Fund VI Co-investment LP"); and 6 First Halder Affidavit, paragraph 10. (ii) Canada Pension Plan Investment Board ("CPPIB"), which represents 20 million contributors/beneficiaries (together the "Buyer Group"); (e) BPEA informed the Company’s Board of Directors (the “Board”) of its interest in an acquisition at a Board meeting on January 13, 2017. The Board established a special committee comprised of two independent directors unaffiliated with either the Buyer Group or the Company’s management (the “Special Committee”). The Special Committee engaged Houlihan Lokey Capital, Inc. (“Houlihan Lokey”), a specialist adviser, to prepare a fairness opinion on the proposed transaction; (f) the Company created a transaction due diligence data room but was not involved in the negotiations between the Special Committee and, inter alia, the Buyer Group between January and April 2017. Discussions took place between the Special Committee’s professional advisers, BPEA and the professional advisers of the Buyer Group; (g) an initial non-binding offer was made by the Buyer Group on April 4, 2017 of US$30 per Share, which offer was rejected. The Buyer Group raised its offer to US$32.00 per Share, but the Special Committee insisted on a “go-shop” provision to permit the pursuit of other offers within a 30 day period. Following further discussions, the Buyer Group increased its offer by 50 cents to US$32.50 per Share; (h) Houlihan Lokey verbally advised the Special Committee that the offer price was fair in advance of providing its formal written fairness opinion; (i) on April 25, 2017, the Special Committee determined that it was advisable to enter into the Merger Agreement, the Plan of Merger and the related transactions contemplated by those agreements. The Board decided to enter into the Merger Agreement on the same date subject to approval at an extraordinary general meeting ("EGM"); (j) the main features of the Merger Agreement were as follows. US$32.50 was the consideration for each Share ("Merger Consideration"). Bach Acquisitions (the “Merger Sub”) would be merged with the Company. The Company would continue to exist as a subsidiary of Bach Finance Limited (the “Purchaser”), but the Merger Sub would cease to exist. The Company would cease to be a listed company and would be beneficially owned by the Buyer Group; (k) also on April 25, 2017, the Merger Agreement and a Share Sale and Support Agreement ("SSSA") were entered into between Premier and the Parent. The Merger Agreement was announced. The previous day, the closing trading price of the Shares was US$27.62; (l) the offer price represented a premium of 17.7% above the pre-announcement closing price and 24.2%, 31.7% and 33.8 % respectively above the weighted average daily trading price of the Shares over the previous 30-day, 60-day and 90-day period; (m) the go-shop period ended 30 days after April 25, 2017 and no other bids were received although Houlihan Lokey on behalf of the Special Committee contacted one potential strategic bidder and 14 potential financial bidders; (n) the EGM was held on August 21, 2017. 79,369,500 Shares (approximately 85% of the shares represented at the EGM) voted in favour of the Merger and 14,006,383 Shares (approximately 15% of the shares represented at the EGM) voted against. The Merger became effective on September 1, 2017 when the Plan of Merger was executed and filed with the Registrar of Companies.

¶ [15] Mr Halder’s Second Affidavit was sworn on April 10, 2019 ("Second Halder Affidavit"). By this time he was no longer Chief Financial Officer of the Company and was residing in England. The Second Halder Affidavit explained the basis on which he had prepared financial projections while in the Company’s employ. In summary: (a) annual budgets prepared by a small team he directed were the primary basis on which the Company made its financial plans. Occasionally projections would be prepared for the purposes of raising debt and/or equity; (b) in August 2016, Mr Halder prepared 5 year projections for the first time as part of the annual budget process. These forecasts formed the basis for all subsequent Company forecasts; (c) in November 2016, Mr Halder prepared an updated version of the August 2016 5 year plan in connection with Project Darwin, another potential sale ("November 2016 Projections"). The 'Bank Model' sheet was not prepared by him but was based on his figures insofar as it included 5 year projections. The additional “Projections to Reach Steady State” covering the period 2022-2026 were neither prepared by Mr Halder nor based on his figures. His own figures for financial years 2017-2022 represented his “ambitious but achievable view at the time”7; (d) the next five year projections prepared by Mr Halder’s team were in April 2017 (“April 2017 Projections”) in the context of the Merger transaction. This update included additional updated material and included additional (presumably electronic) features (the option of turning on and off certain new developments)8. These projections were supplied to Houlihan Lokey. Mr Halder “did not consciously apply any risk-weighting to the April 2017 Projections. They were ambitious”9; (e) after the Merger Agreement was consummated, the Buyer Group prepared its own projections which were more aggressive than the Company’s as part of the Group’s debt raising and syndication process. The main difference was that BPEA models assumed a positive impact on future earnings attributable to Operational Excellence. Based on Mr Halder’s own experience in the business he regarded the BPEA assumptions as being “very ambitious... [but] not outside the realms of possibility”10; (f) between April and August, Mr Halder’s focus was on preparing a full budget. Ian Johnson and the China Bilingual team were requested to prepare more detailed projections because the China Bilingual project was a major area targeted for future development. Existing projections were based on the experience of one such school already operating in China. Between July 2 and August 1, 2017 Mr Johnson forwarded to Mr Halder revised projections for the China Bilingual project. The July 26, 2017 Projections for China Bilingual were far more conservative than those included in the April 2017 Projections. Because Mr Halder did not appreciate the significance of the distinction at the time, he did not include the new figures in the August 2017 5 year plan (the “August 2017 Projections”). Apart from this oversight, the August 2017 Projections represented Mr Halder’s “most up-to-date view of the Company’s foreseeable performance as of the valuation date”11; 7 Second Halder Affidavit, paragraph 12. 8 In the course of the trial, various figures in Excel spreadsheets were digitally altered to demonstrate the financial impact of different inputs on projected share values. 9 Second Halder Affidavit, paragraph 15. 10 Ibid, paragraph 19. 11 Ibid, paragraph 23. (g) Mr Halder deposed that his projections were never risk-weighted in part because risks were difficult to model and in part because the Company operated on the basis that its core business was sufficiently robust that any adverse risks could be absorbed. He also noted that no provision had been made for possible depreciation in foreign currencies, because foreign currency earnings had been converted into US dollars. Alan Kelsey Overview of witness

¶ [16] Mr Alan Kelsey was at all material times Chairman of the Company and, as an independent director, one of the two members of the Special Committee which recommended to the Board of the Company that the Company could properly enter into the Merger Agreement. He obtained a BA and MA from Oxford University. He joined the Company’s Board in 2003 and became Chairman in 2005. He left the Company at the end of August 2017, by which time he had a business career which spanned nearly 50 years.

¶ [17] I found Mr Kelsey to be a credible witness who generally gave his evidence in a careful and (in the face of vigorous cross-examination) impressively objective manner. He acknowledged that he had no experience of merger transactions and deferred to specialist advisers Houlihan Lokey and Kirkland & Ellis as regards the detailed management of the process. He also admitted that a key player in the Buyer Group was a longstanding colleague who might have sought to exploit their personal relationship. For instance: “Q. Didn't you see the inherent risk that, even unconsciously, your personal business relationship with Mr Hennessy might compromise your ability to represent the unaffiliated shareholders fearlessly and effectively? A. No, I had thought about that and I didn't think that and I felt -- I took comfort from the fact I had my own legal and investment banking advisers to give me advice, and I suppose subconsciously, I felt if I'm going to stray off the straight and narrow, they would put me right."12

¶ [18] A mature man with a somewhat Churchillian air, he imperiously swatted away the suggestion that he had in personal terms been improperly motivated in approving the transaction, under cross-examination by Mr Millett QC: "Q. So can we take it from that, Mr Kelsey, that the benefit to you of this merger was that you would not only get $32.50 in cash for your Nord shares as at or after 31 August, but you would also get that price in cash for your restricted units? A. Yes. Q. Was that something of an incentive to you personally to get this deal done? A. At the highest price, yes. But if -- I believe Professor Gompers is postulating a value of $76 and if it had been $76, I would have got 10 million more or something like that. Q. Never mind about what Professor Gompers does or doesn't say, Mr Kelsey, I think the answer is that it was an incentive to you personally to get the deal done. A. It was an incentive for me to get the highest price."13

¶ [19] Although it was never positively asserted that Mr Kelsey had deliberately misled the Court, it was contended by way of submission that the Court should infer from his evidence that the transaction process was seriously flawed. Overview of evidence-in-chief

¶ [20] Mr Kelsey’s First Affidavit was sworn on April 4, 2019 ("First Kelsey Affidavit"). Its main purpose was to explain his involvement in the Special Committee’s work in relation to recommending the Company’s entry into the Merger. He deposed in summary as follows: (a) the work of the Special Committee was primarily carried out by its advisers and he did not personally have a detailed recollection of all that occurred. However, the process is described in detail in the Company’s 12 Transcript Day 3, page 141 lines 1-11. 13 Transcript Day 4 page 98 line 18-page 99 line 8. proxy statement dated July 11, 2017 (the “Proxy Statement”) and the minutes of various meetings and calls; (b) both he and Mr Nicholas Baird (a former British Ambassador to Turkey and Board member since 2015) were appointed as Special Committee members in January 2017. Neither had in-depth experience of mergers, but they each understood the Company’s business well. They considered they were well supported by Houlihan Lokey and attorneys Kirkland & Ellis; (c) both Messrs Kelsey and Baird understood that their role was to negotiate and recommend a deal which was in the best interests of the shareholders in circumstances where Premier was controlled by BPEA and had the voting power to approve any merger. As far as Mr Kelsey himself was concerned: “…To me, this meant that the Special Committee had a particular responsibility to negotiate the best terms possible for the non-BPEA affiliated shareholders (which I refer to as the ‘minority shareholders’) and only to approve such a deal and recommend it to the Board and to shareholders if were satisfied that we had met this obligation. That is the mind-set that I took into this task and which I was confident we had achieved in the result we attained”14; (d) Mr Kelsey summarised his personal reasons for approving the transaction as follows: (1) Premier’s interest in obtaining a high price, which he took for granted, was aligned with the interests of the minority shareholders, (2) the Special Committee was able to negotiate freely with CPPIB as the lead investor, (3) his job was to negotiate vigorously for the best possible deal for the Company and all its shareholders which was likely to succeed, (4) the Special Committee’s experienced team of advisors helped it to negotiate a price which valued the Company fairly and was favourable to minority shareholders; 14 First Kelsey Affidavit, paragraph 9. (5) the merger price of US$32.50 was higher than the traded share price had ever been and Houlihan Lokey advised that it was a fair price from a financial point of view; (6) if another party had been willing to offer better terms, the agreement with the Buyer Group could have been terminated.

¶ [21] A secondary function of the First Kelsey Affidavit was to respond to Experts’ questions. These questions related to, inter alia, the Company’s relationship with BPEA (or Baring), any common interests on the sell and buy sides, why Project Darwin failed, the appointment of Special Committee members and conflicts of interest, the appointment of advisors, the negotiation of the deal price, why the deal was not first offered to a wider pool of investors, the terms of the Merger Agreement (including the dropping of the ‘majority of the minority’ clause and the ‘go-shop’ provisions).

¶ [22] Mr Kelsey’s Second Affidavit was sworn on July 17, 2019 ("Second Kelsey Affidavit) in response to the First Affidavit of Anil Seetharam filed on behalf of the 18th, 27th and 28th Respondents (the “Stockbridge Dissenters”). The essence of Mr Kelsey’s more detailed response to criticisms of the ‘go-shop’ process is captured in the following portion of his written testimony: "...I was and am unaware of there being any process failures with the go-shop process and I do not accept that the go-shop process was just window-dressing as Mr Seetharam tries to suggest. In fact, from my perspective as a member of the Special Committee, I considered the go-shop process to be competently and professionally managed...Had an offer been forthcoming from Stockbridge, the Special Committee would certainly have taken it seriously and considered whether it met the requirements to be taken forward."15 Ian Johnson Overview of witness

¶ [23] Mr Ian Johnson was at all material times Commercial Director of the Company’s China Bilingual programme, who prepared the July 2017 projections which painted a less optimistic picture of this programme’s prospects than was reflected in the Company’s April 2017 and August 2017 financial projections. A Chartered Accountant, he joined the Company in 1995. He left the Company’s full-time employ on August 31, 2019 and 15 Second Kelsey Affidavit, paragraphs 9-10. continued to work from England on a part-time basis, supporting the programme leader Ms Jian Tang.

¶ [24] I found Mr Johnson to be a credible witness who generally gave his evidence in a straightforward and fair manner. Overview of evidence-in-chief

¶ [25] The First Affidavit of Ian Johnson was sworn on April 5, 2019 (the “First Johnson Affidavit”). It was intended to supplement explanations given at the Management Meeting about the China Bilingual programme. His written evidence may be summarised as follows: (a) in China, the Company operated (1) schools for mostly foreign nationals and (2) China Bilingual schools for mostly Chinese nationals; (b) although there were four types of competitor for the China Bilingual business, a greater difficulty was securing land on which to operate the schools. Accordingly, it had not been possible to replicate the initial success of the “NACIS” school; (c) additional challenges included obtaining licenses, construction time and the lack of brand familiarity for Nord Anglia in cities where no international schools already existed; (d) in May 2017 Mr Andrew Fitzmaurice and Mr Graeme Halder asked Ms Tang and Mr Johnson to prepare a 5 year forecast for the China Bilingual business. The forecast model was completed on July 26, 2017. Mr Johnson described the basis for this model as follows: “…In preparing this model we had taken into account recent market research from Parthenon which gave details of the potential for the bilingual schools market in various tier 1, tier 2 and tier 3 cities, the competition which existed, the student numbers and growth in existing schools, the fee rates in those schools and other matters as well as input from our newly established business development team in Shanghai. Ms Tang and I looked at these findings, in order to establish the assumptions for the model...Between us we determined the inputs for student numbers and fee levels and this was developed into a model…”16; 16 First Johnson Affidavit, paragraphs 19-20. (c) the result of this analysis was projections far lower than the previous China Bilingual projections which had been based solely on the performance of the NACIS school in Shanghai. Account had to be taken of the fact that fee increases required regulatory approval. The lower forecast projections were also because, inter alia, the proposed schools were to be set up in smaller, less wealthy and more remote cities, resulting in likely far lower full time enrolment ("FTE") targets being achieved; (f) a significant commercial problem unique to China was the problem of extracting cash from both international and bilingual schools, described as the "non-fungible cash generated in China" problem. Regulatory changes in 2016 were the source of this impediment to extracting distributable profits. Patrick Cordes Overview of witness

¶ [26] Mr. Patrick Cordes was at all material times Chief Financial Officer of Baring Private Equity Asia ("BPEA"). By the date of the trial he was an employee of more than 13 years' standing who had also been a Managing Director since 2015. After graduating from Lehigh University in Pennsylvania in 1997, he had worked for some nine years with Deloitte in New York and Hong Kong. In 2006 he became Financial Controller for BPEA; and he was promoted to Chief Financial Officer in 2008. Mr Cordes was primarily cross-examined about the dual role played by Baring on the "Sell Side" and the "Buy Side".

¶ [27] Mr Cordes was by far the youngest of the Company's fact witnesses. He had seemingly not previously given evidence in a court and at one juncture became somewhat discombobulated by the barrage of questions he was required to answer. He appeared to be a "bright young spark" comfortably embedded in the environment he worked in and with a firm grasp of the relevant facts. I found him to be a credible witness who gave his evidence in a straightforward and fair manner overall. On the contentious issue of the efficacy of the transaction process, for instance, he gave what appeared at first blush to be a very frank response to robust cross-examination by Mr Millett QC: "Q. Who was representing the sellers -- other than the special committee for the unaffiliated shareholders, who was representing the BPEA affiliated funds on the seller's side? A. Technically, we had taken the team and separated it, meaning Kosmo Kalliarekos was serving the sell side," Jack Hennessy was doing buy. But at the same time, it's not -- you know, there is no Chinese wall. You know, investment committee meetings, people are all present. So it's just something to have some consideration on sides. Q. Yes. So is it right to say that actually there was no negotiation between the seller side -- the affiliated BPEA funds on the seller side and those on the buyer side? A. Because, like I said, if anything, it was -- you had CPPIB, who was reticent about price, believing it's too high. So, in fact, we had to work to get them up. So it wasn't that Jack Hennessy was only trying to work down, we actually were working the buyer group up to the level that we believed was the full fair price."17

¶ [28] The Dissenters did not suggest that he was not an honest witness but submitted that Mr Cordes "was not a very impressive witness and the Court should be very cautious before relying on anything he said"18. Overview of evidence-in-chief

¶ [29] The First Affidavit of Patrick Cordes was sworn on July 11, 2019 (the “First Cordes Affidavit”). He was part of the team that structured the Merger transaction and sat in on various meetings where the transaction was discussed. The main function of his First Affidavit was to reply to the First Seetharam Affidavit. His evidence may be summarised as follows: (a) the BPEA Funds which were relevant to the First Seetharam Affidavit were Funds III and IV, which had beneficial interests in the Company pre-Merger, and Fund VI which acquired a beneficial interest through the Merger. Each Fund had separate General Partners; (b) BPEA was a subsidiary of Baring Private Equity Asia Group Limited ("BPEAG"), which was either the parent of or investment advisor to seven Funds established by BPEA from 1997. The deponent referred to BPEA, BPEAG, the Funds and their General Partners collectively as "BPEA Group"; 17 Transcript Day 4 page 143 lines 4-24. 18 Dissenters’ Written Closing Submissions, paragraph 43. “12. BPEA is a subsidiary of Baring Private Equity Asia Group Limited ("BPEAG"), the investment advisor or parent to the investment advisor of all Baring Private Equity Asia funds (collectively 'BPEA Group')”; (c) he further deposed that: “17...the BPEA Group, Fund III and Fund IV were strongly incentivised to achieve the highest possible price for the Company shares beneficially owned by Funds III and IV"; (d) Fund VI only acquired an interest post-Merger via the Merger process along with CPPIB. Fund III beneficially owned a 6.8% stake in the Company and Fund IV 63.7%; (e) although the Proxy Statement showed the Company’s ownership post-Merger would be CPPIB 61.9% and Baring Filing Persons 38.1%, it was always intended to reduce that stake by selling on to co-investors. After the process of syndication which actually occurred, CPPIB only beneficially owned 36.9% and Fund VI beneficially owned 21.1% with co-investors beneficially owning 39.8% and the BPEA Group itself only owning 2.2%; (f) prior to the Merger, the BPEA Group had owned 1.8% so there was not any significant increase in its stake resulting from the transaction. Moreover, BPEA was very mindful of its fiduciary duties to the investors in Funds III and IV, whose interests lay in achieving the maximum possible price. In any event, the Advisory Boards of those Funds approved the Merger Consideration; (g) the threshold for the Baring General Partners to receive 20% of any sale proceeds received by Funds III and IV had been reached, so BPEA itself was incentivised on the Sell Side. There were no hidden benefits accruing to BPEA on the Buy Side. THE DISSENTERS’ FACT WITNESS Anil Seetharam Overview of witness

¶ [30] Mr Anil Seetharam was at all material times Managing Director of the Stockbridge Dissenters. The main function of his evidence was to explain why the Stockbridge Dissenters were, as long-term minority investors in the Company, dissatisfied with the value assigned to their Shares.

¶ [31] Mr Seetharam appeared to me, somewhat like Mr Cordes, also to be a “bright young spark”, with a firm grasp of the private equity world he works in and the facts relevant to his testimony. Despite his obvious enthusiasm about advancing his company’s cause I found him overall to be a credible witness who generally gave his evidence in a straightforward manner. Under cross-examination by Lord Grabiner QC, for instance, he fairly made the following concession: “Q. Anyway, I think you are saying that you can't recall -- well, correct me if I am wrong, you can't recall an occasion where you were asking for information which you reasonably thought you were entitled to but they wouldn't give it to you? A. We asked for lots of things and I would say the majority of times in most public companies, the answer you get is, 'No, we can't give that to you'. Q. What about Nord though? I'm not interested -- A. Nord was no different.”19

¶ [32] His credibility was challenged as regards one aspect of his testimony in particular, namely his evidence that the Stockbridge Dissenters only prepared written valuations of the Shares after the Merger Agreement had been consummated. I shall return to this issue in more detail below. Overview of evidence-in-chief

¶ [33] The First Affidavit of Anil Seetharam was sworn on February 15, 2019 (the “First Seetharam Affidavit”). It explained Stockbridge’s role as an investor in the Company, why it dissented and expressed concerns about the fairness of the go-shop process. Mr Seetharam’s evidence may be summarised as follows: 19 Transcript Day 5 page 29 lines 12-21. (a) Stockbridge first invested in the Company’s Shares in July 2014 and increased its stake until the Merger. It was a long-term investor, not an appraisal arbitrageur, and one of the largest outside investors, holding 6,049,035 Shares at the Merger Date; (b) over that three year period, Stockbridge maintained a constructive dialogue with BPEA’s lead Board representative (Kosmas Kalliarekos), spoke to Company Management and visited school campuses in Hong Kong, Shanghai, Beijing and Houston; (c) Stockbridge strongly believed the Transaction Price seriously undervalued the Company, and made this position known to the Company as soon as the Merger was announced on April 25, 2017; (d) Stockbridge expressed an interest in making a joint bid to representatives of CPPIB on April 28, 2017 and expressed an interest in participating in the go-shop process to Houlihan Lokey on May 3, 2017. It did not pursue this option because it was told any bid would have to be based on publicly available information. Although after the Go-Shop period expired and before the Proxy had been belatedly filed Houlihan Lokey said an unsolicited bid could still be made, this was not viewed as a serious proposition in light of the informational advantages enjoyed by the Buyer Group and the 2/3rds majority interest of BPEA; (e) this left Stockbridge (and presumably other Dissenters) with the sole option of dissenting as they believed there was a significant gap between the Merger Consideration and fair value.

¶ [34] The Second Affidavit of Anil Seetharam was sworn on May 28, 2019 (the “Second Seetharam Affidavit”). It dealt with Dissenter discovery issues. THE COMPANY’S EXPERT EVIDENCE Professor Daniel Fischel Overview of Witness

¶ [35] Professor Daniel Fischel is President and Chairman of Compass Lexecon, a consultancy, with specialities including Valuation and Financial Analysis. He qualified as a lawyer in or about 1979 and in 1982 became a professor of law at Northwestern University School of Law. He was the Lee and Breena Freeman Professor of Law and Business at the University of Chicago Law School between 1984 and 2005 (with his chair being awarded in 1984). Professor Fischel has given expert evidence in numerous cases in the United States for over 40 years. Unsurprisingly, he has published extensively. He has in recent years been accepted as an expert in several appraisal cases in courts including the Court of Chancery of Delaware, which court has a broadly similar appraisal regime to section 238 of the Companies Law.

¶ [36] Although obviously an eminent expert and an impressive witness in general terms, Professor Fischel from time to time gave what appeared to me to be obtuse answers to questions on potentially contentious issues and to be unwilling to countenance any modification of the opinions recorded in his Expert Reports. For instance, when questioned about academic literature suggesting that “go-shops” were transaction devices which might push the merger price down unless they were properly structured, he seemed reluctant to admit even as a matter of general principle that the form a “go shop” took might be even potentially relevant to the resultant merger price: “Q...Let's see if we can cut this short. There is an article you cite yourself by Subramanian and Zhou. It's {F/86/46}. I'm sorry to dive in in the middle of a complex piece of academia, which is on this subject, but do you recognise this? A. Yes, I'm very familiar with this article and I actually discuss it at some length in my report. Q. You do, which is why I'm assuming you are familiar with it. It says in the last paragraph: "We agree with Vice Chancellor Laster's formulation of existing Delaware doctrine." He's got four bands about the sale process: 'As a matter of policy, we further believe that this is a sound formulation, because it encourages good deal processes. The key question, of course, is what deals should go in each band. 'On that key question, we are not suggesting that go-shops are categorically bad, ie automatically pushing the deal into Band 3 or Band 4. There may be good reasons for sellers to want to limit pre-signing competition (eg, to avoid leaks, grab the 'bird in hand', et cetera). However, the key doctrinal takeaway from this Article is that the use of a go-shop in lieu of a traditional pre-signing market canvass (particularly a 'pure' go-shop) should generally push the deal down on this spectrum, but this presumption of a downward nudge can be overcome by a properly structured go-shop." Do you agree with that? A. I don't know what's meant by "a properly structured go-shop". What I discuss in my report is, even with respect to this article, which probably in the spectrum of academic literature on go-shops is more critical of go-shops than many other articles that I also discuss -- but even with respect to this article, it discusses the number of topping bids that have resulted in transactions with go-shops, so I would say that, even based on the commentary in this article, there is not a basis to conclude that go-shops are useless or shams -- Q. No -- A. -- and don't encourage the existence of topping bids. Q. I haven't suggested that and nor do the learned authors of this article..."20

¶ [37] However, the following day, dealing with the same broad topic of transaction process, he gave a more straightforward and fair answer to a question of general principle: "Q... As a general question, though, it's right, isn't it, that whether negotiated sales are better than auctions at producing value for sellers will vary from case to case depending on the facts and circumstances? A. Yes, that's correct."21

¶ [38] And under further cross-examination by Mr Millett QC on the "go-shop" question, Professor Fischel was ultimately willing to make a concession about the limited parameters of his Report: "Q. All right. Let's turn to a different topic. Do you accept that the robustness of a go-shop doesn't depend 20 Transcript Day 6 page 198 line 18-page 200 line 16. 21 Transcript Day 7 page 1 lines 20-24. solely on the legal terms of the go-shop but also the way in which it was conducted? A. Yes, of course, I would agree. Q. And it's right, isn't it, in the way you deal with this, apart from the proxy statement and the contractual documentation, you don't refer to a single contemporaneous document on the record as part of your report, which tends to show how, in fact, the go-shop exercise was conducted? A. That may very well be true, I don't know. Q. So, in fact, you were considering this go-shop entirely in the abstract? A. Well, I not only discussed the articles that you referred to about go-shops, I also looked at the empirical evidence about go-shops, including in the exhibit that you just showed me, but I did not conduct a detailed factual investigation of how this particular go-shop was conducted, other than what's already stated in my reports.22

¶ [39] He was in general terms a credible and objective expert witness. The Dissenters sought to paint the Company’s Expert as “an evangelist for the market price as the primary, and in this case only, data point for fair value”23. That argument deserves careful consideration. It was further submitted that Professor Fischel gave his evidence in a “combative” manner, as an "advocate for the Company...inconsistent with the duties of an expert witness before this court"24. This argument is summarily rejected as an unjustified characterisation of the manner in which the Company’s Expert testified overall.

¶ [40] Nonetheless Professor Fischel did not appear to me to make a single concession on the merits of any issue which was significant to his main opinions in the present case. In these circumstances I consider that the most critical aspects of his testimony should be approached with some care. 22 Day 7 page 17 line 6-page 18 line 1. 23 'Written Closing Submissions of the Dissenting Shareholders', paragraph 11. 24 Ibid, paragraph 20. Overview of evidence-in-chief

¶ [41] Professor Fischel’s Expert Report was dated July 18, 2019. In paragraph 22 of his Report, he summarised his principal conclusions as follows: “ • Market evidence generally provides the best estimate of the fair market value of a company’s common stock because market transactions reflect what willing buyers paid willing sellers for that stock. The closing price of Nord Anglia Education’s ordinary shares on the day before the Announcement Date of $27.42 (the ‘Unaffected Price’) provides a reliable estimate of the fair market value of a Share if market participants believed that there was a possibility that the Company would have been acquired at a premium. When adjusted to account for interim changes in market and industry conditions, the Unaffected Price implies that the fair value of a Share as of the Valuation Date was $30.45. • The Merger Consideration of $32.50 per Share that the Buyer Consortium agreed to pay, and Nord Anglia Education agreed to accept, is another form of market evidence that is relevant to an assessment of the value of a Share. However, the value of the Merger Consideration likely exceeds the fair value of a Share because the Merger Consideration presumably reflects a sharing of the gains the parties anticipated would be created by the Transactions. This conclusion is supported by the process that led to the Transactions, the actions of insiders, the results of the go-shop process, analysts’ target prices, and analysts’ reaction to the announcement of the Transactions. • Based on a DCF analysis of the Company’s contemporaneous projections, the fair value of a Share ranged from $28.64 to $41.19 on the Valuation Date, before adjusting for any discount that may be applicable to minority holdings. For various reasons detailed herein, however, I have concluded that a DCF analysis should be given little or no weight in the determination of the fair value of a Share. • The comparable company multiples analyses performed by analysts, and the financial advisor to CPPIB are consistent with my opinion that the Unaffected Price provides a reliable estimate of the fair value of a Share as of the Announcement Date. However, for reasons detailed herein, I have concluded that a comparable company multiples analysis should be given little or no weight in the determination of the value of a Share."

¶ [42] Further to the Experts’ Joint Memorandum dated August 29, 2019, Professor Fischel prepared a Supplemental Report dated September 24, 2019. A summary of the main conclusions reached in the Supplemental Report is as follows: "5. Professor Gompers’ conclusion that the fair value of the Dissenters’ Shares is $77.45 per Share is implausible on its face because that amount bears no relation either to prices that willing buyers paid willing sellers for the Shares or to the value of the Merger Consideration that the parties to the Transactions reached in arm’s-length negotiations. Moreover, none of Professor Gompers’ reasons for disregarding market evidence withstand scrutiny. Furthermore, Professor Gompers’ DCF analysis is fundamentally flawed, unreliable and fails the reasonableness check that he himself proposed but did not perform. Finally, Professor Gompers’ analysis of the value of control is flawed... 7. Section VI of this supplemental report describes three revisions to my DCF analysis, which, as revised, results in an estimated equity value per Share of $28.34 to $41.32 on the Valuation date, depending on the discount rate. This range contains my estimate of the fair value of Dissenters’ Shares on the Valuation Date of $30.45 per Share. However, for the reasons discussed in my initial report and §IV.A infra, it is my opinion that DCF analysis should have little or no weight in the determination of the fair value of Dissenters’ Shares."

¶ [43] In Professor Fischel’s oral examination-in-chief, he formally produced updated versions of Exhibit S-5 and S-8 to his Supplemental Report and produced four further Exhibits. The revisions reduced the lower point DCF valuation range by 30 cents per Share and increased the upper point in the range by 13 cents. This was a reduction of roughly 1.05% of the initial bottom figure in his range and an increase of roughly 0.31%. As trifling as these changes seem to be, it is noteworthy that they are minor changes to the upper and lower figures in a more generous range of potential values with a gap of just under $13 between top and bottom. THE DISSENTERS’ EXPERT EVIDENCE Professor Paul Gompers Overview of witness

¶ [44] Professor Paul Gompers has since 2000 been the Eugene Holman Professor of Business Administration at Harvard Business School. After graduating from Harvard College in 1987 and obtaining a Masters from Oxford University in 1989, he obtained a PhD in Business Economics from Harvard University in 1993. His pre-2000 academic career includes University of Chicago (1993-1995) and Harvard Business School (1995-2000). Unsurprisingly, he has an impressive list of publications. Professor Gompers has provided expert evidence to various United States courts over, inter alia, the last 5 years, including appearing as an expert in appraisal litigation before the Delaware Court of Chancery.

¶ [45] Applying the standards of British common law courts, it was surprising that Professor Gompers as an experienced expert witness often appeared to me to give his testimony on hotly contested issues in what for me seemed an overly partisan manner. This may partly be attributable to legal cultural differences and partly personality. Professor Gompers’ resume reveals that he is a marathon runner who in his youth was an alternate member of the US Olympic team. The Dissenters’ Expert’s engagement with Mr Boulton QC, who displayed admirable mastery of the expert materials in the course of his cross-examination spanning three days, often resembled an Olympian intellectual contest. A man with stellar academic credentials, Professor Gompers was an expert with strong views which were typically forcefully expressed. He was always clear (if not always convincing), but his initial reluctance to accept hypothetical scenarios which were inconsistent with his clients’ cause diluted any semblance of objectivity on his part. Inevitably, this increased the need for his evidence to be approached with considerable care. Viewing his evidence fairly overall, I would reject the submission advanced in closing arguments that Professor Gompers fundamentally failed to discharge his duties as an independent expert witness altogether.

¶ [46] The following exchange reflected the adversarial and somewhat bombastic stance that Professor Gompers often exhibited under cross-examination. The topic under discussion was four different ways of making adjustments to financial projections to take into account potentially negative outcomes: “Q. And if you didn't like any of those, you might add a premium to the discount rate to reflect the fact that your projections didn't build in the possibility of bad outcomes. A. I would fail you in finance if you did that. There is no basis to do that. That's absolutely wrong. Q. It's sometimes criticised, isn't it, as being a fudge factor, I think is the Brealey and Myers term for it. A. Sir, if I might, I think the best way to do it is, when you are projecting out capacity on schools you would buy or schools that you would build, you would build in what you think historically has been your average across all the schools. So let's just say hypothetically -- and at Nord it's roughly 80 per cent, but some are above and some are below, so projecting those new schools or projecting those schools at that capacity would take into account that some are going to be above and some would be below. So that is, I would think, how you would do it. But at the end, the way you would really want to assess, sir, is to assess how Nord had done historically relative to, in aggregate, they projected. Q. Can I be absolutely clear here, that you say that adding a premium to the discount rate would lead to a fail in finance? A. Correct."25

¶ [47] Mr Boulton QC returned to the same topic later the same day by reference to the Expert's own writings: "Q. If we go to page 3, {F/82/3}, what you are doing here is helping your readers understand how you might account for specific types of risk, diversifiable, firm and country-specific risks. A. Correct. Q. And if we pick it up in the largest paragraph on the page, you talk about the possibility that management of a company steals the assets of the firm away and you say, how can you deal with that. You say you've got to factor it into the cashflows. One method: '... would be to create two scenarios that are probability weighted ...' That would be, on my example, 80 per cent and 20 per cent. And you go on to say: 'It is always possible, however, to adjust the discount rate in such a way to achieve exactly the same 25 Transcript Day 9 page 37 line 25-page 39 line 2. valuation as the one reached by adjusting the cashflows with this weighted scenario.' Do you see that? A. I do. Q. Do you stand by that as being the two ways that you can do it? A. That is correct. So mathematically, if you have a constant per period probability that a country-specific risk factor materialises, that you can derive the maths to show that that would be the case. And so one way, for example, to interpret these country risk premia, which come from Damodoran, is that it's equivalent to assessing whether it's stealing the company, whether it's regulatory imposition, whether it's restrictions on cash, as we were talking about earlier. These are country-specific things. That sort of says that this accounts for that as these constant per period probabilities. Mathematically, it's identically the same thing. Q. I think on the next page, please, operator, {F/82/4}, we can see that you work through an example, a specific example in the penultimate paragraph, and you are considering a risk that a firm will go out of business and you say you can deal with it through the cashflows but you can also deal with it through the discount rate, which is the final paragraph on the page. A. That's correct. This is just deriving the math Q. Yes. So if we could look at page 37 of today's transcript, {Day9/37:7}. I was going through four ways you could deal with a bad outcome, and one of them was you could specifically model a bad outcome or you could do a general provision or you could model different scenarios. That's at {Day9/37:21}. That's one of the two approaches that you take in your article. And the final question I put to you is: {Day9/37:25} 'And if you didn't like any of those, you might add a premium to the discount rate to reflect the fact that your projections didn't build in the possibility of bad outcomes.' Do you see that? A. I do. Q. And you said: 'I would fail you in finance if you did that.' A. And I still would. Q. But, in essence, that's dealing with exactly the same situation that you're writing about in your article, isn't it? A. No, actually I would ask you to -- and if you want to pull a copy from Harvard Business School publishing, if you pull my note on valuation in private equity, you will see a discussion exactly of this. So one of the things, sir, that private equity professionals and venture capital investors often do is they increase their discount rates to 25, 30 or 50 per cent. In that note, sir, you will see that I discuss why that's wrong and why you shouldn't do it. They do it because they think they are adjusting for the risk. And the real question is, if you do that, you are missing all of the important underlying assumptions that go into a DCF, as well as a whole other set of issues. So generally speaking, sir, if I had a student who was looking at a project and they goosed up the discount rate by 20 or 30 per cent, that student would fail. Q. Indeed, Professor Gompers, but nothing about adding 20 or 30 per cent to the discount rate was in my question to you earlier. It was the principle that you have two ways of dealing with this. One is through the cashflows and the other, in certain circumstances, is through the discount rate...."26

¶ [48] The Dissenters’ Expert had an elaborate explanation as to why he believed it was in all circumstances, save for dealing with country risks, inappropriate to provide for risks in financial projections by adjusting the discount rate. However, it seemed obvious that he had responded to the initial very general question in an unjustifiably broad and combative manner. This was not a fair representation of Professor Gompers’ approach overall. He also gave very straightforward answers to some questions about the key (and contentious) elements of his DCF model. For instance: "Just simply want to demonstrate how much money on your valuation turns on this, Professor. Your long-term growth rate here is, I think you say, 1.5 times the weighted average inflation rate. That's correct; yes? 26 Transcript Day 9 page 166 line 12-page 169 line 21. A. Yes. Q. So if we remove the 1.5 times multiplier and simply use the weighted average inflation rate, your 4.23 per cent would go down to 2.82 per cent. Is that right? A. Correct. Q. Let's insert that at B45, simply to see the sensitivity that arises from that assumption. 2.82 per cent, please, operator. $54.10. So something like $21 of value arises simply from an assumption that the company for ever more will be able to increase its tuition fees at 1.5 times the rate. I think that may be an underestimate. If we could go to exhibit 3B, and we have a separate long-term growth rate here for China Bilingual. So on the same assumption, we will change B93 to 3 per cent, please, operator. That's changing long-term growth rate from 4.5 per cent to 3 per cent. Thank you. Let's go back to exhibit 3A. Diluted price per share is now down to $49.71. So a third of the value in your DCF model arises solely from your assumption that from ten years' time forever the company will be able to increase its tuition fees in dollars by 1.5 times the rate of inflation. A. That is correct under the conservative assumptions of no expansion of capacity or acquisitions after the terminal period, but the math is what it is."27

¶ [49] Professor Gompers also admitted that the most value-relevant criticism he made in his Report of the Houlihan Lokey valuation related to the fact that their projections were limited to a five year period. He further admitted that the documentation evidencing the work that they did (obtained by the Dissenters pursuant to section 1782 applications in the United States) "would be helpful in assessing the work"28. Without expressly admitting that the Houlihan Lokey work product was not subject to any criticism, the Court was left with the distinct impression that had it been subject to serious criticism the work carried out would have been put to the sword by Professor Gompers in a detailed and reasoned manner. Instead the position appeared to be that either the Expert's support team had not seen fit to draw the supporting documentation to his attention or (improbably) he had failed to review documents which were obviously relevant to advancing a reasoned critique of Houlihan Lokey's DCF analysis: 27 Transcript Day 10 page 123 line 20-page 124 line 124. 28 Transcript Day 10 page 225 lines 20-21. "Q. So it appears that, despite the time and cost of the 1782 application, that you, in the end, chose not to rely on any of the 2,900 documents? A. I would have to go back and see which documents were produced to me, but they are certainly not relied upon in my first report. Q. And I don't have an equivalent appendix to your supplemental report. But if one searches that document for Houlihan Lokey references, take it from me, Professor, that all we have are a document relating to the go-shop process and the financial model, which we will find at footnote 379 to your second report, that's {HSD/6/112}. A. There is appendix 3 in the supplemental report at {E/85/1} and going forward. Q. It's not up on the court bundle. A. I'm sorry, sir. Q. That's not your fault. Are you aware, Professor, that Houlihan Lokey did, in fact, perform internal analyses that extended the April 2017 projections to reach a steady state? A. They did not utilise them in their valuation in the fairness opinion. Q. That wasn't my question. Are you aware, Professor, that Houlihan Lokey did, in fact, perform internal analyses that extended the April 2017 projections to reach a steady state? A. I don't recall seeing such projections... Q. You can see that these projections -- if we blow up the middle of the page, please, operator -- have been extended out to 2026, can't we? A. Again, I don't know whether or not -- again, not having the underlying spreadsheets for how they calculated this, I don't know if they did the proper thing, which was to take the schools that weren't in capacity and put -- get them up to capacity. So again, I have no way to evaluate how they did the five years of extension in this model."29 29 Transcript Day 10 page 226 line 23-page 227 line 25; Transcript Day 10 page 228 line 9-18

¶ [50] This exchange took place at the end of Day 10 and the second day of the Dissenters’ Expert’s testimony. Thereafter, on Days 11 and 12, it appeared to me that Professor Gompers generally responded to questions put to him in cross-examination in a somewhat more balanced and less combative manner. It remains to consider the following arguments set out in the Company’s Closing Submissions: “127 A particularly telling exchange about how Professor Gompers approaches his expert assignments is recorded at {Day11/178:18} - {Day11/179:23}. In response to the question: ‘Well, yes, you've relied on other class certification reports, but you yourself have never filed a report that concluded that a market was semi-strong efficient.’ Professor Gompers answered: ‘Because I've only been asked to work for the defendants in matters in class certification. It's not in their interests to pay me to write that report, so even once I find it, I either won't write a report or I would write about other issues that I find might be relevant to class certification. I've clearly worked with plaintiffs on other types of matters but, for whatever reason, plaintiffs in securities class action have never approached me to work for them.’ 128 That response is a blatant contradiction of the duties imposed on experts by the FSD Users Guide, particularly B5.2(b)(i), (ii), (iii), (iv), (vii). In ordinary circumstances one might be able to excuse Professor Gompers' statement as a Freudian slip partway through a long cross-examination. Sadly, on the present facts such an excuse is not credible. Professor Gompers made the statement on his second attempt at Mr Boulton QC's question. The statement is consistent with pointed criticism of Professor Gompers in the United States courts. Most damningly it aligns with Professor Gompers' performance in this case where he took a clearly partisan approach, refused to make sensible and obvious concessions and misrepresented key evidence and academic authorities. 129 In light of the above, the Company respectfully submits to the Court that the evidence of Professor Gompers cannot be relied upon when determining the fair value of the Dissenters' shares.”

¶ [51] In my judgment the answer relied upon does not, properly understood, support a finding that Professor Gompers has breached his fundamental duties as an expert in the present case. As Mr Bompas QC pointed out in his oral closing submissions, the role of an expert in a class certification case is fundamentally different to that of an expert at the trial of an appraisal action: "My Lord, Monroe. That case is at authorities bundle 31.1, {AB/31.1/1}, and we covered it in our closing at paragraphs 157 and 158. My Lord, the Monroe case was used mainly as a vehicle for an attack on Professor Gompers' credibility. In our submission, that was unjustified, and there is, you will note from my learned friend's closing submissions, the fact they have chosen at paragraph 128 to say, somehow, that, because of what Professor Gompers said to you about the way in which he was routinely instructed in those kinds of cases, somehow he has not complied with his obligations under the FSD experts' declaration. That is an odd point, to put it at its most polite. But what he says in the passage that's seized upon by my learned friends and quoted -- it needs to be understood in the context of certification procedures in fraud on the market actions in the United States. What happens, as I'm instructed -- and you can see this from all these cases, if you are so minded to take the time to burrow through them -- is that the claimants serve an experts report, the defendants choose whether or not to answer it. It's sequential. There is no burden on the defendant to show efficiency. So if the defendants' expert thinks he can't undermine the claimants' case on efficiency, he won't write a report or he'll write on something else, a different Cammer factor, liquidity, the number of analysts covering the stock. There are four others to choose from. That's all he was saying. And although it's correct that the court in that case in Atlanta rejected Professor Gompers' opinion on market efficiency, it did so in the context -- the very limited context, of a challenge to the rebuttable presumption of market efficiency for the limited purpose of a preliminary application to determine the constitution of a plaintiff class."30

¶ [52] More fundamentally still, the suggestion that Professor Gompers would in any legal context indicate to a client seeking a desired opinion that he was unable on the facts of that case to provide it is on its face entirely consistent with the proposition that Professor Gompers understands his duties of independence to this Court. Overview of evidence –in-chief

¶ [53] The Expert Report of Paul A. Gompers was dated July 18, 2019. The Dissenters’ Expert’s key conclusions may be gleaned from the following portions of that Report: “12. As an economic matter, the price paid for an asset in an arm’s length negotiation between informed parties may be viewed as a reliable indicator of fair value, particularly if that price results from an open, competitive sales process designed to attract multiple potential buyers who are given equal opportunity to bid on that asset. Where an open and competitive process does not exist, it cannot be assumed that the transaction price reflects fair value. 13...I understand that the question of the extent to which Stockbridge was able to participate in the go-shop process is an issue between the parties but that this is not an issue for the expert to resolve. However, I consider that it is relevant in this regard to note that my analysis described in Section VII does not support the contention that the Take-Private Transaction resulted from an arm’s length process nor one that was competitive with multiple bidders... 15. The price at which a company’s stock trades may provide evidence of the company’s value as a going concern... 16....the evidence here shows that Company insiders, including Baring, had access to relevant private information of the Company that was not available to other market participants. Thus the market price cannot be taken to be reflective of the Company’s fair value. 17. In addition, my analyses show low liquidity of the Company’s shares (which I establish empirically, and which was acknowledged by the Company itself in public filings), and a lack of response of the trading price to the disclosure of earnings-related information. Under these circumstances, I conclude that Nord Anglia’s market price prior to and at the time of the Take-Private Transaction cannot be taken to reflect the Company’s fair value... 19. Given that the so-called market indicators of value ...cannot be taken as reliable evidence of fair value, I calculate the fair value of the Dissenters shares using DCF methodology. One of the central tenets of modern finance is that the intrinsic value of a firm is equal to the discounted value of its expected after-tax cash flows... 20. The calculation of fair value in this case is well-suited for the DCF approach, among other reasons, because there were a number of reliable cash flow projections available for the Company, including projections prepared and approved by the Company’s management... 26. Based on my DCF analysis, using the midpoint of the valuation implied by the August 2017 Projections (projections prepared by Company management and closest in date to the Valuation Date) and the valuation implied by the Bank Model, I find that the fair value of the Company’s shares equalled $77.35 per share as of the Valuation Date. 27. While my valuation represents a significant premium over the pre-announcement market and Take-Private Transaction prices, in my opinion this difference can be reconciled based on differences in the information available to the market and to Company insiders, (e.g., Baring) as well as based on the particular features of the deal process, such as absence of competitive bidding and inability of potential bidders to access internal information about the Company.”

¶ [54] Further to the Experts’ Joint Memorandum dated August 29, 2019, Professor Gompers prepared a Supplemental Report dated September 24, 2019. A summary of the main conclusions reached in the Supplemental Report is as follows: “5...My calculation in the Gompers Report yielded a fair value of the Dissenters shares [of] $77.35 per share as of the Valuation Date. I have further revised my calculation in light of discussions with Professor Fischel in drafting of the Joint Report, whereby Professor Fischel and I agreed to use common values for certain inputs in our cost of capital calculations in order to narrow the scope of differences between the experts, as well as to make an adjustment for stock-based compensation. This yields a value of $76.51 per share as of the Valuation Date... 8. My conclusions regarding Nord Anglia’s fair value expressed in the Gompers Report remain unchanged with the exception of the adjustment described above. It is also my opinion that Professor Fischel’s analysis of fair value significantly understates the fair value of Dissenters’ Nord Anglia shares, as described in the Gompers Supplemental Report, and that correcting for even some of the problems in his analysis would lead one to conclude that fair value of the Dissenters’ shares was substantially higher than the Take-Private Transaction Price and closer to my valuation of $76.51 per share."

¶ [55] As a result of agreed input changes, Professor Gompers reduced his initial valuation by 84 cents (or approximately 1.08%). Like Professor Fischel, he was only willing to agree that minimal adjustments were required to his initial DCF analysis. Unlike Professor Fischel, he was not willing to explicitly acknowledge that an appropriate DCF valuation might fall within a range of potential valuations rather than being definitively reflected in a specific sum. JOINT MEMORANDUM BETWEEN FISCHEL AND GOMPERS Agreed Matters Preliminary Matters

¶ [56] It was common ground that the valuation opinions should be expressed as at the Valuation Date on a per share basis. Each Expert also understood that “fair value” meant the value of a Share of the Company operated as a going concern and excluding any advantages which might accrue as a result of the Merger. Valuation Approaches Considered

¶ [57] Each Expert considered the same four valuation methodologies, although they disagreed as to whether the Market Price or DCF approach was more appropriate in the circumstances of the present case. Market Efficiency

¶ [58] It was agreed that the stock market is not generally considered to be “strong-form efficient” i.e. with market prices reflecting both public and private information. A market is “semi-strong form efficient” if the market fully reflects all public information. It was agreed that in a semi-strong efficient market: (a) stock prices respond to and rapidly incorporate new value-relevant publicly available information (market, industry and/or company-specific); and (b) no significant changes in stock price would be expected on days when there was no new value-relevant information. Event Studies

¶ [59] It was agreed that the event study methodology is an accepted means of testing semi-strong market efficiency and that a properly designed event study tests whether a company’s stock price has reacted to company-specific news to a statistically significant extent. The effect of market or industry factors on stock returns is calculated by regression analyses using either a market index (a ‘one-factor model’) or both market and industry indices (a ‘two-factor model’). Relevance of trading volume

¶ [60] It was agreed that a relatively high trading volume can be an indicator that newly available information is being incorporated into a company’s stock price and low trading volumes can raise concerns that such incorporation is not occurring and that the stock is not semi-strong efficient. Comparable Companies and Precedent Transaction Analyses

¶ [61] Neither Expert relied on these analyses. DCF Analysis

¶ [62] The Experts were agreed on the following matters: (a) the general mechanics required; (b) the appropriateness in general terms of using the Company’s August 2017 Projections; (c) the need to modify those Projections (by extension) to reflect a transition to a steady state and to correct minor errors; and (d) the discount rate can be calculated using the Company’s weighted average cost of capital ("WACC").

¶ [63] As regards the elements of WACC, the Experts agreed that: (a) the country risk premium (if any) should be 1.17%; (b) recent academic literature cast doubt on the validity of using a size premium; (c) the equity risk premium should be 5.985%; (d) the risk-free rate should be 2.52% (based on the 20 year U.S. Treasury bond; (e) although they used different market indices and capital structure assumptions to calculate the Company’s equity beta, they agreed that these choices did not account for any material differences that their respective beta calculations yielded; and (f) the depreciation and amortisation assumptions should be consistent with the August 2017 Projections. Main areas of disagreement between the Experts

¶ [64] The Joint Memorandum lists almost 30 “major” areas of disagreement between the Experts. Based on the way the parties conducted the case at trial31, I would distil those areas of disagreement into the following headline issues: (a) whether fair value was adequately reflected in the Market Price or whether a DCF analysis was required; and (assuming a DCF analysis was required); (b) what adjustments should be made to the August 2017 Projections as regards the China Bilingual Project; (c) what adjustments should be made to the August 2017 Projections as regards exchange rates in relation to foreign currency earnings; and (d) how the appropriate discount rate or “beta” should be determined and, having regard to materiality, the following sub-issues: (1) the length of the estimation period, (2) the use of weekly data, (3) the cost of debt, and (4) the terminal growth rate. 31 See e.g. ‘Company’s Closing Written Submissions’, paragraph 110 et seq, and paragraph 130 et seq ‘Written Closing Submissions of the Dissenting Shareholders’, paragraph 282 et seq, paragraphs 336 et seq, paragraph 392 et seq and paragraph 406 et seq. LEGAL FINDINGS: PRINCIPLES GOVERNING THE SELECTION OF THE APPROPRIATE VALUATION METHODOLOGY Cayman Islands authorities

¶ [65] The pivotal statutory provision to the present case is section 238 (1) of the Companies Law which provides as follows: “(1) A member of a constituent company incorporated under this Law shall be entitled to payment of the fair value of his shares upon dissenting from a merger or consolidation.”

¶ [66] The Dissenters centrally argued: “70. To summarise the legal position, in assessing ‘fair value’ what the Court is aiming at is not merely the market price of its shares but rather their value, taking into account all value-relevant information whether or not it was available to the market at the time, leaving out of account any advantages or disadvantages accruing to the Company arising from the Merger Transaction itself. The usual way of arriving at such a value, and one which forms a part of the Court’s assessment of “fair value” in all prior s.238 cases, is through a DCF analysis.”32 [Emphasis added]

¶ [67] Rather than considering which valuation approach should be adopted based on the contending views of the Experts viewed in light of the isolated facts of this case, it seems logical to me to first consider how this question has been resolved by this Court in previous cases. Is there a governing legal principle or does the choice of valuation methodology turn on purely factual considerations? The previous section 238 cases that have gone to trial appear to number only three: Re Integra Group [2016 (1) CILR 192] (Jones J); Re Shanda Games Limited, FSD 14 of 2016 (NSJ), Judgment dated April 25, 2017 (unreported) (Segal J); and Re Qunar Cayman Islands Limited, FSD No 76 of 2017, Judgment dated May 13, 2019 (Parker J) (unreported).

¶ [68] In Re Integra Group, Jones J described the transaction as being “a management buyout...structured as a statutory merger” (paragraph 5). The petitioner contended that the market price of the listed shares reflected fair value and the respondents contended for a DCF analysis combined with (applying a 25% weighting) a market-based comparables approach. These positions were advanced through expert witnesses. Jones J concluded, having considered the expert evidence: “38. The mere fact that a company’s shares are listed on a major stock exchange, in this case the LSE, does not lead to the conclusion that a valuation 32 *Written Opening Submissions for Trial on Behalf of the Dissenter Groups*. methodology based upon its publicly traded prices is necessarily the most reliable approach towards determining its fair value for the purposes of a section 238 court appraisal."

¶ [69] The presumed free float was 67.5% of the issued capital and the trading volume of that free float was 16% and 9% in the last two years before the merger. The trading volume was dramatically lower and the median bid-ask spread marginally higher than the corresponding figures for comparable companies. Based on his view of the expert evidence, Jones J rejected the full-blown market price approach contended for by the petitioner. The shares were valued at $11.70 per share rather than the offer price of $10.

¶ [70] The transaction in Re Shanda Games was described by Segal J as a "take private transaction led by the principal shareholders and management of Shanda" (paragraph 6). That shareholder indirectly controlled approximately 69% of the company's voting shares; the Second Buyer Group controlled just over 90% of the voting power (paragraphs 34, 50). Both experts agreed on the DCF approach. The offer price was $7 per ADS; the petitioner’s expert valued the shares at $7.10 per ADS and the respondents’ expert valued the shares at $27.03 per ADS, reflecting what Segal J described as a "fundamentally different view" (paragraph 66). Segal J recorded helpful findings on the general approach to expert evidence in the section 238 appraisal context. Although Segal J’s views were expressed in the context of a dispute as to how a DCF analysis should be applied rather than whether it is appropriate to use the methodology at all, they provide valuable general guidance on the section 238 judicial function: "85. I note, and find persuasive, the comments of Vice Chancellor Strine in Andaloro v PFPC Worldwide Inc, (Andaloro) Court of Chancery, Delaware, New Castle 2005 Del. Ch. Lexis 125 at [34]. In making the fair value determination, the court may look to the opinions advanced by the parties' experts, select one party's expert opinion as a framework, fashion its own framework, or adopt piecemeal, some portion of an expert's model methodology or mathematical calculations. But, the court may not adopt an 'either-or' approach and must use its judgment and an independent valuation exercise to reach its conclusion.' More recently in his opinion in Re Appraisal of Dell Inc 2016 WL 3186538 (Dell) Vice Chancellor Laster quoted with approval dicta in this issue from a number of other cases as follows...33: 33 At pages 42-43. "In discharging its statutory mandate, the Court of Chancery has discretion to select one of the parties' valuation models as its general framework or to fashion its own". M.G. Bancorporation, 737 A.2d at 525-26. "The Court may evaluate the valuation opinions submitted by the parties, select the most representative analysis, and then make appropriate adjustments to the resulting valuation". The court also may "make its own independent valuation calculation by . . . adapting or blending the factual assumptions of the parties' experts." M.G. Bancorporation, 737 A.2d at 524. It is also "entirely proper for the Court of Chancery to adopt any one expert's model, methodology, and mathematical calculations, in toto, if that valuation is supported by credible evidence and withstands a critical judicial analysis on the record." Id. at 526. "When . . . none of the parties establishes a valuation that is persuasive, the Court must make a determination based on its own analysis."

¶ [71] Instructive approaches adopted by Segal J to the expert evidence included averaging the beta figures proposed by the experts and averaging the experts’ proposed terminal growth rates (over a terminal period following a 5 year transitional period). These were approaches which subsequently received the imprimatur of the Court of Appeal: In re Shanda Games [2018 (1) CILR 352] at paragraphs 75 and 92 (Martin JA).

¶ [72] Re Qunar, FSD No 76 of 2017 (RJP), Judgment dated May 2, 2019 (Parker J) (unreported), is the third trial judgment in a section 238 case. The company in that case had its shares listed on NASDAQ and the merger was a take-private transaction supported by the majority shareholders. As here, a special committee recommended the transaction to the board of directors and engaged Duff & Phelps who opined that the negotiated price of $30.39 per ADS was a fair one. This was midway in their DCF range which implied values of between $26.59 and $34.52 per ADS. The company’s expert’s valuation was based 50% on a DCF valuation and 50% on a market price analysis and contended for $28.09 per ADS applying a 4.7% minority discount. The dissenters’ expert contended for $125.96 per ADS, using a DCF analysis alone and applying no minority discount. Re Qunar’s factual matrix has important similarities with that in the present case and accordingly Parker J’s judgment provides valuable guidance as to how issues such as, inter alia, the following should be approached: (a) when a DCF valuation approach alone should be preferred to a market price approach; (b) the inferences to be drawn from a huge gap between a DCF valuation and the market price; (c) the determination of when a market is sufficiently efficient to enable the market price to provide reliable evidence as to fair value; and (d) the correct approach to a DCF computation as regards matters such as assessing the reliability of management projections, the calculation of the appropriate beta, whether a ‘Blume adjustment is required’ and the appropriate terminal growth rate.

¶ [73] Parker J unsurprisingly described the gap between the valuations as “very considerable” (paragraph 16). He concluded that a 100% DCF approach was not warranted because the market was not shown to be inefficient and the transaction was conducted fairly. However, the merger process and merger price did not apparently play a central role in the case. The majority shareholder held 95% of the shares, but this did not mean in and of itself that the market in the shares was inefficient. He found that the market price “can reasonably be relied upon as good evidence of value. It therefore also provides a good cross check against the DCF outcome of fair value” (paragraph 141). The blended approach contended for by the company’s expert was approved but the minority discount was valued at nil. Parker J also approved the dissenters’ expert’s China-based terminal growth rate of 4.75%. He left the experts to revise their valuations in light of his findings on various disputed issues, which included the various elements of the appropriate discount rate. It seems self-evident that the ultimate result must have been far closer to the petitioning company’s expert’s valuation than to the dissenters’ valuation. The most significant general findings made by Parker J which have potential persuasive weight for the purposes of the present case were the following: “408. The huge difference to the financial outcome of fair value in this case as contended for by the parties, is attributable primarily to the reliance by the Dissenters and Mr Osborne solely on a DCF method of calculation. This method has given rise to a number of issues where, in addition to the arguments concerning the Management Projections, the reliability of models, comparators, assumptions and the validity inputs have been tested. It has also given rise to the consideration of much academic and practitioner literature and financial analysis. 409. It has produced, on the Dissenters’ case, a valuation which is way off the market price. It is, as Mr Osborne accepted, not credible once the systemic undervaluation theory of shares in companies with their operations in the PRC, but which are listed on US exchanges, has been rejected. Such a result would mean that investors and others who work in the US markets had significantly underestimated the Company’s value prior to the Merger and has wider implications for other businesses with similar operations...”

¶ [74] Just over a month after I reserved judgment in the present case, the Judicial Committee of the Privy Council delivered its judgment in Re Shanda Games [2020] UKPC 2. As regards the approach to interest, not an issue in controversy at the present stage, Lady Arden summarised the Judicial Committee’s views as follows: “57. The judge exercised his discretion on the rate of interest that should be paid by Shanda in a manner which on its face was unassailable on appeal. He followed the practice in Delaware. He took the midway point between a rate of interest representing the return on the unpaid appraisal monies that a prudent investor could have made and the rate of interest that the company would have had to pay to borrow the equivalent sum.”

¶ [75] As regards the minority discount issue, of potential relevance to the present case but not addressed by either Expert, the Privy Council held that in principle a minority discount was applicable to a valuation under section 238 of the Companies Law. Lady Arden summarised the Judicial Committee’s conclusions on this issue as follows: “55. It follows that the judge should not have held that fair value always means no minority discount (see, for example, judgment of the judge, para 93, second sentence). That could not be a bright-line rule to be applied in every case. Similarly, it was not open to CICA to hold that a minority discount should invariably be applied as a matter of law. The legislature’s direction is to find the “fair value” of the dissenter’s shareholding. Because of the narrow scope of this appeal, the Board is not in a position to rule out the possibility that there might be a case where a minority discount was inappropriate due to the particular valuation exercise under consideration.”

¶ [76] It is noteworthy that the question before the Court was not what value should be assigned to the relevant shares, but whether section 238 mandated a pro rata valuation or not: “56. As explained, the only issue which the parties have argued is whether the shares of the dissenters should be valued on a pro rata basis or not. The parties have not sought to argue that the value should be something other than CICA found it to be if section 238 does not require a pro rata valuation of the dissenters’ shares...”

¶ [77] It is helpful to compare some general features of the present case with those of the previous section 238 cases. Compared with Integra, where there was seemingly no controlling shareholder at all and a ‘free float’ of 67%, here Baring and affiliates beneficially owned 67% of the Company’s Shares pre-Merger reflecting a free float of roughly 33%. Professor Gompers opined without apparent contradiction that the weekly trading volume of the Company’s Shares was just over 0.5% of all stock (over the 5 years up to and including 2017), far lower than for comparable companies34. Measured in relation to the one third of all Shares reflected by the free float that means roughly 1.5% of that float traded weekly. In Integra, trading volumes were also comparatively low. In Shanda Games, the size of the free float was apparently comparable to the present case, roughly 31%. It appears that the market in the shares in the latter case was far less liquid than in the present case. In contrast, the free float in Qunar where the controlling stake was 95% was much smaller and the trading volumes presumably even lower.

¶ [78] There is no precedent for placing primary or sole reliance on the market price in any of the three previous section 238 cases which went to trial. Delaware authorities

¶ [79] The potential relevance of Delaware and/or Canadian appraisal jurisprudence to the application of section 238 of the Companies Law is now well settled. The fact that section 238 “was heavily influenced by the Delaware and Canadian law” was seemingly first judicially acknowledged by Jones J in In re Integra Group [2016 (1) CILR 192] at paragraph 20. As noted above, Segal J found Delaware jurisprudence in appraisal litigation to be of persuasive value in Re Shanda Games Limited, FSD 14 of 2016 (NSJ), Judgment dated April 25, 2017 (unreported) (at paragraph 156 et seq). In Re Qunar FSD No 76 of 2017 (RJP), Judgment dated May 2, 2019, Parker J observed: “34. The jurisprudence in Cayman is relatively young in comparison to Delaware and Canada and guidance from those jurisdictions has been found to be helpful by the Cayman courts in terms of the approach to the similar issues the courts of those jurisdictions have adopted. However, the approach is not always to be followed in Cayman as there are differences in the language of the relevant legislation, the policy behind it, insofar as one can identify that, and procedure. 35. I have considered in this case the principles from, in particular, the Delaware jurisprudence in the same way as Segal J and the Court of Appeal 34 Expert Report, paragraph 201; Appendix VIII-5. did in Shanda, but I have also had regard to, for example, English cases concerned with solving problems relating to the concept of fair value, albeit in different statutory contexts, as did the CICA in Shanda."

¶ [80] The most authoritative judicial support for the relevance of, in particular, Delaware appraisal case law at the date of the trial was to be found in recent Cayman Islands Court of Appeal decisions. Re Shanda Games [2018 (1) CILR 352] was a case where the Court of Appeal directly held that the Delaware approach to computing interest should be followed but the Delaware approach to minority discounts should not be followed. The more general reliance placed by Segal J on Delaware jurisprudence was not doubted. In the present case, the general relevance of Delaware case law was not in dispute, as regards the general approach to appraisal cases. In re Qunar [2018 (1) CILR 199] was a case where the Court of Appeal was guided by the Delaware approach to dissenter discovery. Rix JA (Field JA and Goldring P concurring) opined as follows: "63. Next I go to the jurisprudence of Delaware, which has had long familiarity with the concept of fair value in the context of dissenting shareholders and a statute with similar wording to s.238...I do not however take any account of Dole35 as constituting any authority or precedent. In any event it comes from a separate jurisdiction. However, I cannot ignore considerations which are addressed in Dole as though such matters did not exist in the world. That would be to shut my eyes to the realities of life. 64. What I find in Dole is a sophisticated, well-informed, modern judgment (delivered in 2014) with extensive citation of jurisprudence...I cite passages in Laster V.C.’s opinion to exemplify the power of its reasoning..."

¶ [81] In my judgment there is (or ought to be) a distinction between the relevance of Delaware authorities on matters of law and procedure and the relevance of such authorities to the more commercially-focussed mechanics of the appraisal process. For the section 238 trial judge at least, seeking to appraise shares which have been listed on a US stock exchange based on expert evidence from witnesses who predominantly give expert testimony in Delaware and other US courts, the commercial and ‘cultural’ elements of US appraisal jurisprudence will likely more often than not be very helpful indeed. The same applies to commentary on the distinctive judicial characteristics of the Court’s appraisal function. After I reserved judgment in the present case, the Privy Council confirmed that the Delaware approach to valuing the shares of a minority (on a pro rata basis as opposed by reference to the value of the actual minority shares) does not apply 35 In re Dole Food Co. Inc. (Appraisal), (2014), 114A.3d 541 (Laster V.C.). to section 238 appraisals: Shanda Games Ltd.-v-Maso Capital Partners Ltd et al [2020] UKPC 2. Lady Arden nevertheless opined as follows: "49. The Cayman Islands courts would still be able to use the jurisprudence of the Delaware and other courts if placed before them as a source of guidance on some particular issue. It goes without saying that the jurisprudence of Delaware is of great value in this field. However, it should also be borne in mind that there may be different rules in related contexts, such as a different regulatory scheme for corporations issuing securities to the public and the duty of fairness owed by a controlling stockholder to other stockholders, which are not found in English and Welsh law. In addition, there may be different economic and social policy considerations affecting legislation in Delaware."

¶ [82] Lord Grabiner QC in his opening oral submissions primarily sought to demonstrate that the Delaware courts now favour a market-based evaluation approach to a DCF analysis. However, a passage counsel cited indirectly helped to illumine (a) the Court’s distinctive commercially-focussed role under section 238, and (b) the fact that US appraisal experts are possibly generally more partisan than a British Commonwealth judge might expect them to be. In Dell, Inc.-v-Magnetar Global Event Driven Master Fund Limited 177A.3d 1 (2017) (Supreme Court of Delaware). Valihura J opined as follows (at paragraph 35): "We pause to note that this appraisal case does not present the classic scenario in which there is reason to suspect that market forces cannot be relied upon to ensure fair treatment of the minority. Under those circumstances, a DCF analysis can provide the court with a helpful data point about the price a sale process would have produced had there been a robust sale process involving willing buyers with thorough information and the time to make a bid. When, by contrast, an appraisal is brought in cases like this where a robust sale process of that kind in fact occurred, the Court of Chancery should be chary about imposing the hazards that always come when a law-trained judge is forced to make a point estimate of fair value based on widely divergent partisan expert testimony." [Emphasis added]

¶ [83] Dell, Inc. helpfully warns trial judges that their legal training does not really equip them to resolve valuation disputes between valuation experts whose estimates are based on the DCF approach and are far apart. Where market indicators such as share or transaction price are reliable indicators of the intrinsic value of the shares, these ultimately straightforward indicators should not be ignored in favour of far more esoteric financial estimates. It is noteworthy, however, that the gap between the merger price and the dissenters’ DCF valuation was a stunning $23 billion in Dell. The company had a “deep public float” with more than 5% of its shares trading each week36. There was no controlling shareholder; Mr Dell owned only 13.9% of the company’s shares37. The merger process was found to be “robust”. Yet the Delaware Supreme Court did not direct the Court of Chancery to simply apply the transaction price; the matter was remitted for reconsideration in light of the following critical findings (at paragraph 37): “Given that we have concluded that the trial court’s key reasons for disregarding the market data were erroneous, and given the obvious lack of credibility of the petitioners’ DCF model—as well as legitimate questions about the reliability of the projections upon which all of the various DCF analyses are based—these factors suggest strong reliance upon the deal price and far less weight, if any, on the DCF analyses.”

¶ [84] The position under Delaware law post-Dell appears to me to be that even when the market price (either based on an efficient market or a robust sales process) is a good indicator of the intrinsic value of a company’s shares, a DCF valuation will at least be taken into account if the relevant management projections about future cash flows are themselves reliable38. This appears to have been the case prior to Dell based upon cases referred to by the Dissenters’ counsel in their opening oral submissions, including In re Petsmart, Inc 2017 WL 2303599, upon which Mr Adkin QC relied. Vice-Chancellor Slights crucially opined as follows (at paragraph 32): “The first key to a reliable DCF analysis is the availability of reliable projections of future expected cash flows, preferably derived from contemporaneous management projections prepared in the ordinary course of business. As this court has determined time and again, if the ‘data inputs used in the model are not reliable,’ then the results of the analysis likewise will lack reliability. And, as the experts in this case both agree, to be reliable, management’s projections should reflect the ‘expected cash flows’ of the company, not merely results that are ‘hoped for’.” 36 Paragraph 7. 37 Paragraph 9. 38 The Company’s counsel forwarded two more recent Delaware decisions to the Court after I had reserved judgment and had almost completed the present Judgment. The Dissenters’ counsel objected. I have not taken these decisions into account.

¶ [85] At the beginning of the Vice-Chancellor’s judgment, however, the following insightful observations were made about the peculiar nature of the judicial appraisal function (at paragraph 1): "I would not be the first to observe that the trial of an appraisal case under the Delaware General Corporation Law presents unique challenges to the judicial factfinder. The petitioner bears a burden of proving the "fair value" of his shares; the respondent bears a burden of proving the 'fair value' of the petitioner's shares; and then the judge, as factfinder, assumes, in effect, a third burden to assign a particular value 'as the most reasonable in light of all of the relevant evidence and based on considerations of fairness.' The role assigned to the trial judge in this process independently to review 'all relevant factors' that may inform the determination of fair value, if not unique, is certainly unusual. It is unusual in the sense that the judge is not bound by the positions on fair value espoused by either of the parties. Indeed, the trial court commits error if it simply chooses one party's position over the other without first assessing the relevant factors on its own. Yet it cannot be overlooked that the judge’s decision in an appraisal case follows a trial—an honest-to-goodness, adversarial trial—where the parties are incented to present their best case, grounded in competent evidence, and to subject their adversary’s evidence to the discerning filter of cross-examination. The trial court then reviews the evidence the parties have placed in the trial record and does its best to 'distill the truth'. In this regard, at least, the appraisal trial is no different from any other trial. The court's determination of 'fair value', while based on 'all relevant factors', must still be tethered to the evidence presented at trial. The appraisal statute is not a license for judicial freestyling beyond the trial record."

¶ [86] In re Appraisal of Petsmart, Inc. is also instructive as regards illustrating what type of auction process has been judicially viewed as robust. The merger process was initiated by an "activist hedge fund" which had acquired a 9.9% stake in the company, not by a controlling shareholder. 27 potential bidders were identified between August 2014 and October 2014, and 15 potential bidders signed non-disclosure agreements. A data room including non-public information was made available to the potential bidders and management made presentations to them. Five bids were received by October 31, 2014; on November 3, 2014, four were allowed to continue further with their bids. The Board in early December was still considering alternatives to a sale. Final bids were considered on December 10, 2014. There were three separate competing bidding groups, and one group expanded its membership near the end of the auction process increase its bidding capacity. The highest bid was ultimately accepted. The merger was approved by over 77% of all shareholders and over 99% over those who voted. The dissenters did not challenge the efficacy of the process as such. The valuation process deployed by the Ad Hoc Committee’s advisers was subjected to apparently strong criticism, as was their impartiality; it was also argued, inter alia, that market conditions limited the range of participants in the auction process, such that the Board did not make a properly informed decision. The Vice-Chancellor crucially concluded: “Importantly, the evidence reveals that the private equity bidders did not know who they were bidding against and whether or not they were competing with strategic bidders. They had every incentive to put their best offer on the table... ...Respondent has carried its burden of demonstrating that the Merger Price of $83 per share was the result of a ‘proper transactional process’ comprised of a robust pre-signing auction in which adequately informed bidders were given every incentive to make their best offer in the midst of a ‘well-functioning market.’” FINDINGS: THE RELEVANCE OF THE MARKET PRICE TO THE FAIR VALUE ASSESSMENT Overview

¶ [87] In my judgment the only real dispute between the Experts as regards the relevance of the Market Price of the Shares is one of degree or emphasis. Professor Fischel opined that the Market Price was the best indicator of fair value, but acknowledged that a DCF analysis was a useful cross-check of the market value. Professor Gompers agreed that the NYSE as a whole constituted a semi-strong efficient market and opined, in effect, that because of MNPI a DCF analysis was in the circumstances more reliable as a guide to the intrinsic value of the Shares. As the Company submitted in its Closing Submissions: “216. At the outset it is important to note that Professor Gompers does not conclude that the market is not semi-strong efficient. He simply says that he has not seen evidence that it is semi-strong efficient.”

¶ [88] Having regard to the way in which this Court and the Delaware courts have previously approached the question of fair value in the cases referred to above, the Court’s appraisal function does not permit me to simply make a choice between the Market Price contended for by Professor Fischel and the DCF valuation contended for by Professor Gompers. Nonetheless in assessing the expert evidence, I cannot ignore the fact that Professor Fischel adopted a more balanced and less absolutist position by explicitly acknowledging that a DCF analysis could properly be taken into account and carrying out his own detailed DCF analysis (which essentially provided a range of potential values). He did this despite contending that the Market Price was the only appropriate valuation method and he further implied in his oral evidence that his DCF analysis could be used by the Court to confirm the appropriateness in general terms of his Market Price figure, if the Court had doubts about the reliability of the Market Price39.

¶ [89] Professor Gompers did not appear to me to even tacitly concede the possibility that his assigned DCF valuation might benefit from downward adjustment in light of the far lower Market Price and Transaction Price figures. In the final substantive paragraph of his Supplemental Report, Professor Gompers opined as follows: "429. My conclusions regarding Nord Anglia’s fair value expressed in the Gompers Report remain unchanged. It is further my opinion that Professor Fischel’s analysis of fair value significantly understates the fair value of Nord Anglia, as described in this Gompers Supplemental Report."

¶ [90] Professor Fischel concluded his Supplemental Report by opining that: "106....the estimated equity value of a Share based on the Adjusted August 2017 Projections ranges from $28.34 to $41.32 on the Valuation Date, depending on the discount rate. This range of estimated equity values per share contains my estimate of fair value of Dissenters’ Shares on the Valuation Date of $30.45 per Share. However, for the reasons discussed in Fischel Report...it is my opinion that DCF analysis should have little or no weight in the determination of the fair value of Dissenters’ Shares."

¶ [91] While all critical opinions must be weighed on their respective merits, I feel bound to approach the most controversial aspects of Professor Gompers’ opinions with considerable caution. Professor Fischel generally appeared to me to approach his valuation task with more objectivity than Professor Gompers. That said, the clear picture painted by the evidence overall points to the following broad conclusion. A DCF analysis should be given considerable weight in the Court’s valuation process, but not to an extent which generates a value which is significantly at variance with the Market Price, viewed together with the Transaction Price. Efficiency of the market in the Company’s Shares

¶ [92] Professor Fischel conducted an Event Study to assess the extent to which the market in the Company’s Shares responded to the publication of value-relevant information. The results showed far greater volatility on News Days than on other days which he opined was consistent with the market being “efficient”40. Appendix C-2 recorded a 4.5% deviation on “News Days” compared with a 1.81% deviation on “Other Days”. He also opined that the “Cammer” requirements for establishing efficiency in the context of securities fraud litigation were met in relation to the market for the Company’s Shares. It was not suggested that any recognised test for establishing “semi-strong form efficiency” had been met.

¶ [93] Professor Gompers conducted an Event Study with the specific aim of discovering whether evidence existed for “semi-strong form efficiency”41, although nothing seems to turn on this terminological difference. In general terms, he set out to measure the same cause and effect relationship to value-relevant information as did Professor Fischel. However, he considered that a 5% variation on News Days (two out of 10 days) was statistically significant. His analysis was based on 10 days when earnings announcements were made; Professor Fischel took into account not just days when earnings reports were published, but other information (Form 6K filings and annual reports) as well. In addition, Professor Gompers opined that it was necessary to analyse whether the price deviation was consistent with what one would expect having regard to the nature of the information published. In his Expert Report, he opined: “182. In an efficient market, the stock price should not only respond to new and value-relevant information, but the direction of response (in terms of the sign of the residual return, i.e. whether it was positive or negative) on each event day should be in line with the expected impact of the news being disclosed to the market.”

¶ [94] Professor Gompers recorded the results of his analysis of this additional criterion in Appendix VIII-3 to his Expert Report. He concluded that “the price movements of Nord Anglia’s stock on earnings announcement days were rarely consistent with the various measures of expected impact of news (or lack thereof)”42.

¶ [95] Professor Fischel criticised the conclusion that a statistically significant variation on two of 10 days did not meet the 5% threshold; but this was a case of apples and oranges. Professor Gompers only treated as significant a variation or “residual” on any day of at least 5%; Professor Fischel in his Supplemental Report applied the 5% standard to both the amount of the variation on any relevant day and to the number of days on which a statistically relevant variation occurred as a percentage of the total number of “News Days”43. When Professor Fischel was cross-examined on Day 6 by Mr Millett QC on his Event Study, attention focussed on the amount of the variation (5% for a one-tail test and 10% for a two-tail test) but not on the import of the number of days on which statistically significant variations occurred. Brief mention was made of the percentage of days on which he found statistically significant variations (roughly one-third) shortly before a short adjournment, but Professor Fischel did not initially suggest that this percentage of days on which statistically significant variations occurred was 40 Expert Report of Daniel Fischel, Appendix C, paragraph 7. 41 Expert Report of Paul Gompers, paragraph 174. 42 Expert Report, paragraph 191. 43 Supplemental Report of Daniel Fischel, paragraph 15. particularly meaningful44. The matter was pursued after the break, and Professor Fischel clearly stated that 8 out of 23 days was statistically significant without meeting counsel’s challenge to explain the precise basis for that opinion45.

¶ [96] In essence, Professor Fischel opined that if there was a statistically significant variation on only one occasion out of 23, this might be attributable to mere chance, but 8 such occurrences were inconsistent with the hypothesis that no value-relevant information was reflected in market activity in relation to the Shares. He also opined one would not expect to see a 100% alignment of statistically relevant price fluctuations on News Days. While I accept this conclusion, I also find by way of partial acceptance of Professor Gompers’ opinions on this issue that neither Event Study provided compelling evidence of market efficiency. I accept that some efficiency was demonstrated by each Expert’s findings, objectively viewed.

¶ [97] Was it relevant to assess whether the market reacted in a way that you would expect the market to react? I find that Professor Gompers was right to suggest that this is a potentially relevant consideration. However, as Professor Fischel convincingly asserted under cross-examination and in his Supplemental Report46, it is not entirely straightforward in all instances to identify what would qualify as a logical reaction to a particular announcement and Professor Gompers’ analysis of this issue is arguably flawed. I do not consider it necessary or possible to resolve this specific difference between the Experts as to whether the statistically relevant price fluctuations also reflected logical responses to the relevant announcements. This controversy ultimately depends on underlying facts which were not adequately explored in evidence.

¶ [98] In my judgment the pivotal consideration is that the statistical evidence generated by both Experts’ Event Studies does demonstrate that there is some evidence of semi-strong form market efficiency, albeit not as compelling as it might be in other cases. As Professor Fischel opined in his oral evidence47: "First of all, I didn't say it's binary. I said I can imagine circumstances where some cases are clearer than others but, generally speaking, there aren't degrees of semi-strong efficiency. That's really what I meant to say." Liquidity of the market in the Company’s Shares

¶ [99] I find that it is ultimately obvious that while the Shares were to some extent actively traded, the free float was not a large one (some 33% of the total Shares) and so the market was not so liquid that the need to look beyond the Market Price does not seriously arise. Professor Fischel appeared to me to exaggerate the weight to be attached to the liquidity factor while Professor Gompers understated it. I consider it to be self- 44 Transcript Day 6 page 52 line 16-page 53 line 21. 45 Transcript Day 6 page 61 line 10-page 63 line 6. 46 At paragraphs 17-21. 47 Transcript Day 6 page 13 lines 13-17. evident that the more heavily shares are traded the more weight that can be attached to the cumulative weight of commercial judgments being made by the buyers and sellers of a particular company’s shares. In Dell, Inc.-v-Magnetar Global Event Driven Master Fund Limited 177A.3d 1 (2017) (Supreme Court of Delaware), the fact that there was a “deep public float” was one of the reasons for criticising the trial judge’s decision to give no consideration at all to the relevance of the market price.

¶ [100] Professor Fischel points out that in the year before the Merger was announced, 1.53% of the public float was traded every week which suggests that the Shares not held by insiders were actively traded by some 98 institutional investors over the year48. Clearly the Shares were not illiquid. But this does not undermine Professor Gompers’ point that the percentage of Shares which were traded was comparatively low49. In my judgment this factor does serve to diminish the weight which might otherwise be attached to the Market Price of the Shares.

¶ [101] As regards the comparatively wide “bid-ask spread” which Professor Gompers suggested could constitute evidence of an impediment to market efficiency, I find no sufficient evidential foundation for concluding that this factor had any material impact on the efficiency of the market in the Company’s Shares.

¶ [102] My critical finding on liquidity is that the proportionate size of the public float in the Company’s Shares was sufficiently small to limit the weight that might otherwise be attached to the Market Price alone as reflective of the fair value of the Shares. However, I also find that the volume of Shares traded was not sufficiently small to raise real doubts about whether or not the relevant market was semi-strong form efficient. MNPI

¶ [103] In the Company’s Closing Submissions, the contending positions on MNPI are summarised as follows: “215...Professor Gompers claims that ‘there is evidence that the market did not have access to all relevant information about Nord’s fair value’ because ‘it did not have access to internal management projections’. ‘As a result’, Professor Gompers claims, ‘Nord Anglia’s stock price would have deviated from the Company's fair value if those projections were value-relevant.’ Professor Fischel demonstrates that Professor Gompers' analysis is flawed for two reasons. First, there is a difference between raw information about a company (such as information about its business, business strategy, and its quarterly operating results, which the Company was obligated to and did disclose) and management projections, because management projections reflect management’s opinions about the company’s prospects, which are not necessarily material. Therefore, even if management projections were not publicly disclosed, that would not establish that ‘the market did not have access to all relevant information about 48 Supplemental Report, paragraph 23. 49 Expert Report of Paul Gompers, paragraphs 201-203. Nord’s fair value’ as Professor Gompers claims. Second, Nord’s management projections were publicly disclosed when the Company filed a preliminary proxy statement on 9 June 2017 {H/414}, and when the Company filed the Proxy Statement on 11 July 2017 {H/444}. However, the Company’s stock price did not increase significantly following either of these disclosures and closed at $32.54 on 21 August 2017. Moreover, the release of management’s projections did not cause analysts to substantially change their assessments of the value of the Company. Both these findings imply that the management projections did not contain material information.”

¶ [104] This summary is helpful because it captures the high point of Professor Gompers’ evidence on MNPI, which relies primarily on the non-publication of Management Projections. What individual actors (be they insiders or outside market analysts) subjectively thought of the market value from time to time I regard as being of minimal relevance for the reasons advanced by Professor Fischel. However I am bound to reject the submission that the Management Projections should be disregarded “because management projections reflect management’s opinions about the company’s prospects, which are not necessarily material”. It is well recognised that reliable future cash flows prepared by management provide a sound basis for a DCF valuation. And in this case the Company’s Mr Halder was personally involved in assisting Baring to raise financing for the Merger in reliance on the relevant Management Projections. It surely does not lie in the Company’s mouth to contend post-Merger that its Projections are not at all material, and the equivocal assertion that they are “not necessarily material” is, perhaps, tacit acknowledgment of this. The Company’s non-materiality case is ultimately, with respect, a somewhat unimpressive one. The Management Projections were indeed published with the Proxy Statement on June 9, 2017. The publication did not result in any discernible market reaction; accordingly, it is submitted, the previously non-public information was not material.

¶ [105] In fairness, both the “not necessarily material” and “not material points” are supported by Professor Fischel’s written evidence50. But based on the Company’s own factual evidence, I find that the November 2016 Projections, the April 2017 Projections and the August 2017 Projections were clearly material to an assessment of the intrinsic value of the Shares in a general sense. These Management Projections, I find contained (to use Professor Fischel’s words) both “raw information about” the Company and “management’s opinions about the company’s prospects”, being opinions based on that raw information. Mr Halder agreed that the Company when retaining Houlihan & Lokey to advise the Special Committee gave the following representation: “The Company further represents and warrants that any financial projections delivered to Houlihan Lokey have been or will be reasonably prepared in good faith on bases reflecting the best currently available estimates and judgments of the future financial results and condition of the Company. The Company will promptly 50 Supplemental Report paragraphs 12-13. notify Houlihan Lokey in writing of any material inaccuracy or misstatement in, or material omission from, any information previously delivered to, or discussed with, Houlihan Lokey."51

¶ [106] Mr Halder testified that his primary task was preparing budgets and making forecasts for future budgeting purposes. The forecasts were clearly based on hard data, not speculative guesswork, although future risks were difficult to predict and short-term forecasts were more reliable than longer term ones. He explained under re-examination by Lord Grabiner as follows: "Q. Which parts of the business specifically, if there are such parts of the business -- which parts of the business would be less predictable? A. So I think the 43 schools, as I say individually, are fairly certain because you know what they are. You then move on to acquisitions and green fields and they are more difficult because there is a degree of uncertainty then in terms of timing and where and how much."52

¶ [107] It appears to be factually undisputed that after the Projections were published on June 9, 2017 (preliminary proxy statement) and July 11, 2017 (Proxy Statement) the Company’s Share price did not increase significantly. However, I do not accept Professor Fischel’s invitation to infer from this fact that the Management Projections did not contain material information which had not previously been published. After all, as Professor Fischel himself opined in a parallel context, it is not always logical to expect a specific market reaction from particular information. If that proposition holds good in the ordinary course of business, it must have even greater force in the uniquely complicated context of the Company with its controlling shareholder on the eve of an EGM to consider and likely approve the Merger Agreement. I do infer that the Projections and related proprietary information did not contain an easily identifiable ‘smoking gun’ which made it obvious that the Merger Price was grossly undervalued. This is entirely consistent with the nature of the Projections as estimates not containing easily digestible “good news” or “bad news”. Rather, the Projections contained material which might be fed into complex modelling designed to generate a clearer prediction of how Management’s judgment as to the Company’s future performance would potentially translate into share value. The April Projections were used by Houlihan Lokey to develop a DCF analysis and by Baring to develop its own models for raising financing for the Merger.

¶ [108] In another context, Professor Fischel convincingly opined that it is not always straightforward to determine what an expected and rational response is to particular 51 Transcript Day 2 pages 58-59. 52 Transcript Day 2 page 214 lines 11-18. items of market information about a listed company. In the present case, the Proxy Statement was filed in circumstances where the market had reasons to expect that the Merger would be approved so that many (if not all) purchasers at this juncture would (by the Company’s own account) be arbitrage investors “betting that by dissenting they will be able either to negotiate a settlement above their share purchase price or to persuade the Court to adopt a DCF analysis reflective of a high valuation, all with the safety net that the likely minimum fair value figure even if the dissent litigation is unsuccessful would not be much lower than the merger price”53. If arbitrage investors were indeed purchasing after the announcement of the Merger Agreement with a view to exploiting dissenter rights, it is not easy to draw any clear inferences as to the impact of the publication of the April 2017 Projections embedded in the Proxy Statement on the Share price in the weeks immediately preceding the EGM. If bets were being placed on the vagaries of appraisal litigation, the commercial context is far removed from the traditional context in which market reactions to public information are assessed. The Experts after all essentially agreed that in the merger context the relevant market price measure (assuming it is appropriate to use the market price to determine share value) is the market price unaffected by knowledge of the Merger Agreement.

¶ [109] The Company also argued that there was no MNPI because it gave regular briefings as required by the NYSE. I summarily reject that argument as the relevant question is not whether there was some disclosure of material information but the converse: was there some material information which was not publicly disclosed. Obviously there was some material information which was not publicly disclosed; the elaborate Highly Sensitive Documents (“HSD”) regime which the Company championed for the present proceedings, not to mention the non-disclosure agreement aspects of the Go-Shop process, are the proof of the pudding in this respect. It matters not that the Dissenters were unable to identify a non-disclosed “smoking gun” which clearly pointed to a value far higher than market value.

¶ [110] In my judgment the relevant threshold question pertinent to deciding whether a DCF valuation should be pursued is whether the existence of MNPI, and the possibility that the Shares were undervalued by the market, justifies looking beyond the Market Price with a view to arriving at a more reliable conclusion as to fair value. As Professor Gompers stated under cross-examination by Mr Boulton QC in explaining why he had not considered the Market Price: “A. It would only be necessary to perform if you could conclude that the market for Nord’s shares were both semi-strong efficient and that there was no material non-public information because, in order for this to be a relevant benchmark or metric, you have to be able to assume that the pre-offer share price reasonably reflected the underlying value of Nord. Absent that, if the market isn't semi-strong efficient and/or if there is material non-public information, this analysis is 53 Company’s Written Opening Submissions for Trial, paragraph 23

¶ [111] A secondary question is to what extent (if any) an analysis of any potentially material non-public information (such as the Projections and the Parthenon Reports) suggests that a higher value should be placed on the Shares than the value reflected in the Market Price or the Transaction Price. I find that the Market Price did not incorporate MNPI which was potentially relevant to fair value. I am unable to go so far as to find, without the cross-check of a proper DCF valuation, MNPI which was actually relevant to fair value was not incorporated. The extent to which this MNPI actually indicates a fair value which is significantly different to the Market Price can only be determined by reference to the DCF analysis which both Experts carried out. Professor Fischel agreed that even if the Market Price was a valid measure of fair value, a DCF analysis was a useful cross-check. If there was quite obviously no potentially relevant MNPI in this case, it makes no sense that Houlihan Lokey should have asked Management to provide information which was not at that point public and conducted its own DCF analysis for the purposes of its fairness opinion. Potential lenders should not have been interested in Mr Halder’s presentations on, inter alia, future cash flows, presentations which apparently revealed internal information about the Company’s past and projected future performance. Baring in preparing its own models for the Buy Side ought to have been content to rely on public market information; instead they were (according to Mr Halder) very interested in the Company’s internal projections. Market analysts ought to have been content to reference the market value of the stock without considering it useful to consider a DCF analysis.

¶ [112] However in my judgment the most compelling evidence of the existence of MNPI which was not disclosed to the market by the Company during the relevant Adjusted Market Price period is the HSD “hullabaloo” at the hearing of the Summons for Directions. I was persuaded by the Company to approve (in part in relation to “valuable proprietary research”) what Mr Steinfeld QC referred to as “a quite draconian protective regime”. Summary: relevance of the Market Price

¶ [113] For these reasons I find that although there is some evidence that the market in the Company’s Shares was semi-strong form efficient, the Market Price does not provide reliable prima facie evidence of the fair value of the Shares because: (a) the evidence of the efficiency of the market in the Company’s Shares is not sufficiently strong to justify appraising the fair value based on the Market Price alone; and (b) further and in any event, there was private information (most significantly Management Projections) which was material to the intrinsic value of the Shares and which key market actors involved in the Merger felt it desirable to take into account. FINDINGS: RELEVANCE OF THE TRANSACTION PRICE Overview

¶ [114] There was a dispute between the Experts and between the factual witnesses as regards the efficacy of the Transaction Process and how reliable the Transaction Price is as a guide to the fair value of the Shares. The Company’s broad position was that the Transaction Price was the product of a genuinely arms’ length and fair process which generated the highest possible price which could be achieved through a sale to an available buyer. The Dissenters’ broad position was that the Transaction Process was seriously flawed and was effectively designed to achieve a sale between related parties which was likely to generate an artificially depressed price. By the end of the trial it was clear that: (a) the Transaction Process had been carried out in good faith in circumstances where the beneficial interests on the Sell Side and the Buy Side had no or no material overlaps and to the extent that Baring was involved on both sides of the transaction, it was incentivised to achieve the best possible price; (b) the Transaction Process was not as robust as it might have been, primarily because Baring was motivated in part by the desire to effect a quick sale and in part by the desire to sell to one of its existing clients (CPPIB) and held (through Premier’s 66.8% shareholding in the Company) what amounted to a blocking vote in relation to approving the Merger Agreement at the EGM and potentially blocking any competing bid; and (c) accordingly, while the Transaction Price could obviously be relied upon to some extent as an indicator of fair value, it was not, standing by itself, a sufficiently reliable cross-check for the Market Price-based valuation.

¶ [115] Mr Kelsey’s straightforward evidence satisfied me that the Transaction Process was carried out by the Company through the agency of an independent Special Committee who relied heavily on reputable financial and legal advice. The Dissenters’ primary case that the entire process was in effect a complete sham tainted by unmitigated conflicts of interest, formulated prior to discovery, was revealed to be little more than a conspiracy theory which was not supported by an objective analysis of the actual facts. Mr Cordes’ evidence satisfied me that there was no significant overlap of beneficial interests on both sides of the Transaction, despite the fact that Baring Funds were indeed involved on both sides of the negotiating table. I find that the Transaction Process was an arms’ length one despite the fact that it was entered into between related parties, was negotiated by human actors who generally worked as part of the same team and also involved a controlling shareholder. Those factors do not discredit the evidential value of the process altogether; they merely diminish the extent to which reliance can be placed on the Transaction Price as indicative of the fair value of the Shares. Mr Cordes in my view described the underlying commercial position honestly when he stated (under cross-examination by Mr Millett QC): "I look at it slightly differently, actually. The way I look at it is there was an intention to sell the interest from Funds III and IV and there was a fiduciary duty to get the highest price and the question then was what do we believe is full fair value and at that level is there a transaction to be done that we would participate in, you know, or -- if someone else, that's fine."56

¶ [116] The Dissenters’ initial suspicions about conflict of interest were justified by the fact that Premier and Management pre-Merger held nearly 70% of the Company’s Shares and the purchaser (Bach Finance Ltd, also controlled by Baring) was acquiring 100% of the Shares. However, Mr Cordes explained that Fund III and Fund IV (the main investors in Premier) were not beneficially involved (either before or after the process of post-Merger syndication) on the Buy Side. Fund VI (with different investors) held a 21.1% interest on the Buy Side and the largest single investor on the Buy Side was CPPIB (admittedly a client of Baring). The BPEA Group itself only beneficially held 2.8%. The true position was not easy to ascertain because of confidentiality obligations owed to the underlying investors, and was hardly obvious as the Proxy Statement affirmed that no change of control would be triggered by the Merger Agreement. The proposed deal was touted in partial reliance on the fact that the change of control rules would not be triggered, which implied the same parties were involved on both sides of the Transaction.

¶ [117] Having accepted the critical assertion that Baring had a fiduciary duty to the beneficial owners of Funds III and IV to maximise the sale price, the suggestion that the negotiated price was an entirely uncommercial one must be rejected. Mr Cordes attended one Advisory Council meeting for Funds III and IV on the Sell Side and one meeting for Fund VI on the Buy Side and prepared minutes of those meetings (which were not produced). He testified that the largest investors in each Fund attended the joint meeting for Funds III and IV and that no one voted against the Merger57. I see no basis for doubting his testimony that the main underlying investors on the Sell Side on April 24, 2017 approved the Merger Agreement and the main Buy Side investor, CPPIB, had not invested on the Sell Side. There was therefore no factual support for the opinion expressed by Professor Gompers in Part VII of his Expert Report that there was no 56 Transcript Day 4 page 142 lines 16-23. 57 Transcript Day 4 page 162 line 15 –page 164 line 19. evidence that the Merger was an arm’s length transaction. His potential reasons “Why the Baring Funds Would Be Willing to sell Below Fair Value”58 were not supported by the factual evidence at trial.

¶ [118] On the other hand Professor Gompers was clearly right, to an extent, to opine that the existence of a controlling shareholder appears to have influenced the Special Committee’s negotiating stance to some extent. It seems obvious that potential outside bidders would have regarded the blocking vote as a potential negative factor, especially since the Buy Side had what the Dissenters’ Expert described as “unlimited matching rights”. The go-shop period was only 30 days; the Merger Agreement omitted a “majority of the minority” voting requirement for the EGM; potential bidders were indeed at an informational disadvantage to the Buy Side. His conclusions on this issue were comparatively modest: “...there is evidence to conclude that, as an economic matter the Take-Private Transaction price can be taken as a reliable indication of the fair value of Nord Anglia. Ultimately, however, whether the deal process as it occurred resulted in the fair value of the Nord Anglia’s shares to be paid to the Company’s shareholders is an empirical question that can be answered by performing a properly executed valuation.”59

¶ [119] Professor Fischel concludes in Part V of his Expert Report that an analysis of the process, viewed in light of the Market Price (and, implicitly, without any need for a DCF valuation), supports a finding that the Transaction Price was higher than fair value. This conclusion is in my judgment not supported by the facts proven at trial which cry out for a DCF analysis to confirm (or as a cross-check for) the true extent to which the Transaction Price does or does not reflect fair value.

¶ [120] I accept Professor Fischel’s more general proposition that the Transaction Price “is another form of market evidence that is relevant to an assessment of the fair value of a Share”60. I find that it carries evidential weight because it does reflect an arm’s length bargain negotiated between loosely related parties through the agency of closely connected human actors but also significantly: (a) the Merger Agreement was concluded in circumstances where at least some attempts were made to find other bidders and none came forward in a serious way; and (b) the Transaction Price offered a premium on the then current market price. 58 Expert Report page 50. 59 Expert Report paragraph 136. 60 Expert Report, paragraph 44. The Transaction was an arm’s length process - but not a “robust” one

¶ [121] The Transaction process was not a “robust sales process” in the sense described in Dell, Inc.-v-Magnetar Global Event Driven Master Fund Limited 177A.3d 1 (2017) (Supreme Court of Delaware) (discussed above). Here there was a controlling shareholder and there were no competing bidders. There was a potential conflict of interest arising from Baring being involved on both sides of the Transaction, but this was addressed through the establishment of the Special Committee. Mr Kelsey was one of the two members and he freely admitted that he had never served on such a committee during his previous career. In answer to a question from Mr Millett QC he described his own qualifications for being Chairman of the Board as follows61: “So what was your particular sphere of expertise that you brought to the board? A. I don't think it was a particular sphere of expertise. It was common sense and the desire to run a board properly.”

¶ [122] It is relevant that the Transaction was not the first and only attempt by Premier to sell its Shares. Mr Kelsey testified that there were numerous approaches from potential purchasers but that Project Darwin in late 2016 was regarded by him as being the most substantive one. He and fellow Company director Mr Hennessy sat on the transaction committee in relation to Project Darwin. He rejected as “far too simplistic” the suggestion that, because Premier controlled appointments to the Board, it in reality controlled the transaction committee itself62. When the same issue was raised again later, Mr Kelsey responded63: “And the board was independent, the special committee was independent. The only way that Barings could do anything about it was to actually replace the board, and to replace the board would have been a major operation and I've no doubt the New York Stock Exchange and the SEC and a few other people would have had a great interest if they had seen that taking place. So I disagree fundamentally with what you are saying.”

¶ [123] Mr Kelsey acknowledged that he had considered with Mr Hennessy, who sat with him on the Project Darwin transaction committee, acting for the Buy Side that there might 61 Day 3 page 51 lines 20-22. 62 Transcript Day 3 page 70 line 23. 63 Transcript Day 3 page 118 lines 12-19. be a risk of their generally close personal relationship influencing his ability to act in the best interests of the independent Shareholders. However, he explained64: "No, I had thought about that and I didn't think that and I felt -- I took comfort from the fact I had my own legal and investment banking advisers to give me advice, and I suppose subconsciously, I felt if I'm going to stray off the straight and narrow, they would put me right."

¶ [124] However it is important to acknowledge the extent to which the negotiated terms of the Merger Agreement did support Premier’s majority position, as Mr Kelsey conceded. An initial draft proposed by the Special Committee included a requirement that a majority of the minority should approve the Transaction at the EGM; this was excised at the insistence of the Buyer Group. A 45 day Go-Shop period was reduced to 30 days. A “force the vote” clause was inserted entitling Premier to call for a vote at an EGM even if the Board changed its recommendation, unless the Company was entitled to terminate the Merger Agreement65. On top of this, under the SSSA, Premier agreed to vote in favour of the Merger Agreement at the EGM. This guarantee of voting support might appear significant to the uninitiated, but Professor Fischel’s evidence demonstrated that this is a common feature in negotiated transactions involving controlling shareholders66.

¶ [125] I nonetheless find that the Special Committee acted in good faith in the best interests of minority Shareholders acting on independent legal and financial advice in approving the proposed Merger. The Transaction was substantially an arm’s length one, albeit one entered into between connected parties on terms that made competing bids from highly motivated parties possible, rather than on terms which positively encouraged competition. Although the Go-Shop period started on April 25, 2017, at the end of the month Houlihan Lokey were still working on a list of potential bidders to contact which Mr Kelsey admitted he was only actioning through an email dated May 3, 2017. However, the contents of the email give an insight into why at the time Mr Kelsey appears to have acknowledged that the Go-Shop process was a somewhat ritualistic undertaking: "Dear Louis, I attach a copy of the list of names that Houlihans are going to contact in the go-shop process. I know some of them are long odds hopes and that anybody really interested will probably make 64 Transcript Day 3 pages 141 lines 6-11 65 Transcript Day 4 page 5 line 18-page 56 line 21. 66 FISCHEL 4: 'Negotiated Transactions with a Controlling Selling Shareholder August 1, 2014 to August 31, 2019'. contact themselves but we want to be sure we have conducted the process thoroughly and 'left no stone unturned...'

¶ [126] In my judgment the admitted informality of the way the obvious conflicts were dealt with by sophisticated actors during a commercially significant negotiation suggests a ‘relaxed’ transparency that would not be manifested if the Transaction was not an honest one. For instance, during Mr Cordes’ cross-examination by Mr Millett QC, the following exchanges took place67: “Q. Yes. So my question again: when did Premier make that commitment? A. The thing is that with Premier and with the buyer entity here, being related, it's not something that requires such a formal process, so in fact, in the real world it's not so formal as you are implying. Q. I see, and that's because they are both on the same side, effectively? A. On opposite sides but in the same team, if you will. Q. Yes, it wasn't arm's length, was it? A. Yes, it was.”

¶ [127] Although the robustness of the entire process was diminished to an extent which was material by the existence of a controlling shareholder which had already agreed to sell to the affiliated bidder, this did not deprive the Transaction of all commercial reality. There was a genuine desire to sell at that time because of the maturity of Funds III and IV. As Mr Cordes stated, “the nine year holding is a long time so the desire to sell was there”68. Although the terms of the Go-Shop process were not as enticing to other bidders as they might have been (not least because Premier had already agreed to vote its Shares in favour of the Merger Agreement at the EGM), there is no credible evidence that any seriously interested bidders willing to pay a substantially higher price were rebuffed. And the Transaction Price was (a) approved by the Special Committee based on credible independent financial advice, and (b) higher than any price the Shares had traded at in the preceding year.

¶ [128] In my judgment Mr Kelsey (who made it clear that Houlihan Lokey, “the experts”, were “running the process”69) appeared to be somewhat confused by legal nuances rather than speaking untruthfully when he disagreed with a portion of the Proxy Statement recording the views of the Special Committee about the unlikelihood of a successful competing bid being accepted in light of Premier’s support for the bid accepted on the terms of the Merger Agreement70. This view was confirmed when Lord Grabiner put 67 Transcript Day 4 page 170 line 19-page 171 line 4. 68 Transcript Day 4 page 188 lines 18-19. 69 Transcript Day 4 page 86 lines 22-24. 70 Transcript Day 4 page 50 line 4- page 53 line 19. the controversial extract from the Proxy Statement in its wider context to the witness in re-examination71. Mr Kelsey’s real disagreement was with the implication that at the time he regarded the Go-Shop mechanism as, in effect, a sham. At the end of this portion of his testimony, he critically explained his clear and coherent practical view of the status of the Go-Shop mechanism as follows: "I go back to what I said many times before, that under the go-shop we were entitled to entertain superior proposals and if we found a superior proposal that was -- that qualified, we would follow that superior proposal."72

¶ [129] Mr Millett’s cross-examination of Mr Kelsey clearly established that the Go-Shop process made it somewhat difficult for competing bids to be made, Premier’s blocking position apart. The Proxy Statement containing detailed information about the Company was only filed after the process ended. Access to ‘inside’ information without negotiating a non-disclosure agreement was problematic. Raising financing within the requisite timeline could be potentially difficult for some interested competing bidders. But whether any of these factors mattered in real commercial terms crucially turns on whether any serious potential bidders were actually deterred from making a superior proposal.

¶ [130] Mr Seetharam attacked the fairness of the process and most pertinently stated in his First Affidavit that Stockbridge did not pursue a bid through the Go-Shop process because it was told it would have to do so based on publicly available information. Although Stockbridge was also told it could later make an unsolicited bid, this was not viewed as a viable option because of the informational advantages enjoyed by the Buyer Group and the controlling position of Premier. On its face, this evidence did not even seek to support a finding that the Go-Shop process operated in a way which actually (as opposed to merely hypothetically) deprived the minority Shareholders of the benefit of a superior competing bid.

¶ [131] Under cross-examination by Lord Grabiner QC, it was admitted that Stockbridge was closely connected with Berkshire Partners LLC and that “together Stockbridge and Berkshire had a first-class understanding of private equity investments... [i]ncluding take-private transactions”73. Mr Seetharam was reluctantly forced to admit that whatever ongoing review of the position may have taken place, after Stockbridge learned of the Merger in April 2017 and considered its options, it never decided to actually make a topping bid74. The strong inference from this evidence is that the Go-Shop process itself had no pivotal impact on Stockbridge’s decision not to make a competing bid. Mr Seetharam was also keen to point out that Stockbridge was a long-term Shareholder, and in this regard creditably conceded that pre-Merger attempts to 71 Transcript Day 4 page 106 line 12–page 108 line 5. 72 Transcript Day 4 page 53 lines 15-19. 73 Transcript Day 5 page 4 line 17-page 5 line 1. 74 Transcript Day 5 pages 16 line 22-page 17 line 4. obtain information about the Company’s affairs had been dealt with by the Company in the same generally restrictive way that other public companies dealt with similar requests75. However I also inferred from this evidence in relation to Stockbridge’s historical connection with the Company that it was in a better position than a complete outsider to form a view as to whether it made sense to make a “topping bid”.

¶ [132] So Stockbridge clearly elected not to make a competing bid, in part at least because its best judgment was that it would be an uphill task to out-bid a deal supported by a controlling shareholder. It also seems somewhat surprising that if the Transaction Price was viewed by Stockbridge as obviously far lower than fair value that it did not demonstrate a greater interest in making a superior bid. Other potential bidders were notified of the Go-Shop process but, as Mr Kelsey suggested at the time, any seriously interested parties would likely have made an unsolicited bid after the Merger Agreement was announced without waiting for the Go-Shop process to start and without waiting for the Special Committee to solicit competing bids. No other potential bidders came forward.

¶ [133] The absence of competing bids is a factor which cuts both ways. I accept that in part the Transaction as consummated in the Merger Agreement broadly viewed did not encourage competing bids from the speculative investor or the faint-hearted. But I equally accept that the absence of competing bids also suggests that there was no deep reservoir of untapped alternative potential purchasers ready and willing to make a purchase. The Special Committee was commercially entitled to form the view that ‘a bird in the hand was worth two in the bush’, assuming of course that the Transaction Price was indeed fair to the minority public Shareholders. The proposed Merger Transaction was in broad principle one Premier was entitled to press for. Mr Kelsey explained why the final offer was accepted rather than risking losing the deal and considering alternative possible transactions in the following logical way: “No, what we -- at that stage we looked at -- we assessed what alternatives there could be, other than accepting an offer, if that offer was deemed to be fair and we wanted to accept it. And the alternatives didn't look attractive, weren't really going to add anything and, in any event, BPH, Barings, whatever you want to call it, could block it, because they were a [majority] shareholder, if they didn't like them.”76 Summary

¶ [134] Accordingly, the Transaction Price does provide some evidence of what price a willing sophisticated seller and a willing sophisticated buyer were willing to exchange for the Company’s issued Shares in the real world, namely cash consideration in excess of US$2 billion77. Moreover, Houlihan Lokey advised that the price was fair. The fact that it was in the 'same ballpark' as the Market Price does suggest to some extent that the Market Price is not a wholly inappropriate indicator of the fair value of the Shares, either as a cross-check for the Market Price or in its own right. The Transaction Price is what the Company itself initially represented was the amount which reflected the fair value of the Shares (including a premium). At first blush, the Transaction Price appeared to me to have much more credibility than the Market Price as an indicator of fair value, most importantly because it was informed by MNPI which was not available to the market. But the entire concatenation of circumstances surrounding the Transaction process still leaves room for anxious doubts about whether it can be relied upon (in part or in whole) without considering the more elaborate DCF analysis in relation to which both Experts, for somewhat different reasons, undertook to assist the Court. FINDINGS: WHAT VALUE CAN BE DERIVED FROM AN APPROPRIATE DCF VALUATION OF THE SHARES? Overview

¶ [135] The question of what weight should be given to a valuation based on a DCF analysis will be considered after first determining what value is indicated by an appropriately reliable DCF analysis. The main competing valuations were the following: (a) Professor Fischel: a range of US$28.04-US$41.45; (b) Professor Gompers: US$76.51; and (c) Houlihan Lokey, independent advisers to the Special Committee: US$27.63 - $39.03 (relied upon by the Company as a fall-back position).

¶ [136] In his closing oral arguments, Mr Boulton QC invited the Court to consider Houlihan Lokey’s work when evaluating the competing DCF analyses: "I do want to spend three or four minutes, my Lord, on Houlihan Lokey because, in my submission, they have not featured as heavily in this hearing as they should have done, given their role. They were -- and this is not contradicted -- the independent financial adviser appointed by the special committee. More than that, they used the April 2017 projections that had been 77 Proxy Statement, page 49. prepared by management and they, therefore, had unfettered access to anything that Professor Gompers identifies as MNPI. We will come on to the slight expansion of that category in Professor Gompers' oral testimony from his written. But your Lordship will recall in his reports he only identifies the projections. Well, Houlihan Lokey have those projections, my Lord. Nobody has suggested they are not independent.... And one sees that from their analysis at {H/337.3/1}, which is the presentation to the special committee, which has been referred to -- this is the version referred to and relied on by Professor Gompers. If we go to page 6, {H/337.3/6}, your Lordship will recognise the football field. And this essentially is the basis by which Houlihan Lokey satisfied themselves that the consideration in this transaction was fair to the holders of shares, and they did that by reference to a DCF that covered a broad range from $27 to $39, rounding down in each case, and also by reference to three selected companies analysis comparables, two of which produced ranges below the transaction consideration and one produced a range that covered the transaction consideration, although was, at its midpoint, somewhat lower78..."

¶ [137] The Dissenters' Written Closing Submissions summarised their position as follows: "465. Finally on this part of the valuation case, there is striking support for the conclusion as to value reached by PG, namely a value of $76.51 per share. 466. This is to be found in the Bach Model which passed back and forth between the Company and its management and Baring during the summer of 2017 (see the numerous iterations of the Bach Model, with email discussion about detailed aspects of the inputs, set out in the Appendix of Bach and Bank models with this Closing. Within the Bach Model not only were there very detailed spreadsheets concerned with all aspects of the Company's business, including of course borrowing and covenant compliance, but there was also a spreadsheet tab containing a DCF valuation. This spreadsheet was designed to allow the user 78 Transcript Day 13 pages 87 lines 9-24; page 89 line 19 –page 90 line 9. to choose between various scenarios. This DCF valuation is referred to by PG at PG1, [360] to [367]377. The valuation would imply a range of per share values at the $80 to $90 mark."

¶ [138] The Experts were agreed in general terms on the appropriateness of using the August 2017 Projections as the foundation for the DCF analysis. Professor Fischel, of course, in principle believed that management estimates of what might happen were less reliable than other market data such as the Market Price and the Transaction Price. However, there was some controversy as to precisely how they should be interpreted. Should they be interpreted, as it were, as "best estimates" on their face? Or should account be taken of Mr Halder's evidence at trial that they were "stretching" estimates which were not risk-adjusted? Ancillary to this question was whether the Company should be able to rely on downwards adjustments to the China Bilingual projections based on work concluded by Ian Johnson in late July but not initially incorporated in the August 2017 Projection, supposedly due to an oversight. In my judgment it is obvious that those adjustments must be taken into account, even though the timing of the adjustments (shortly before presumably huge sums had been borrowed on the strength of more optimistic projections) was at first blush somewhat eyebrow-raising.

¶ [139] As noted above, the following headline issues were controversial at trial: (a) what adjustments should be made to the August 2017 Projections as regards the China Bilingual Project and as regards exchange rates in relation to foreign currency earnings; (b) the appropriate terminal growth rate; and (c) how the appropriate discount rate or "beta" should be determined and, having regard to materiality, the following sub-issues: (1) the use of weekly data, (2) the need for a 'Blume' adjustment, and (3) the cost of debt. The reliability of the August 2017 Projections

¶ [140] In the Dissenters' Written Closing Submissions, it was argued that Mr Halder "parroted the party line on a number of issues, most notably, projections. For example, where he was questioned on the forecasts, he was at pains to repeat that the mantra that these were “ambitious but achievable” though, tellingly, he said that he couldn’t remember “the exact person who came up with the [phrase]” {Day2/190:9-14}79. In the next paragraph, the following point was made: “36. The Court is invited to have regard to the numerous contemporaneous documents which contradict Mr Halder in which it was suggested to third parties that the projections represented ‘best estimates’. That is what he himself told the SC on 21 April 2017 – that the April projections were management’s ‘current best good faith estimates of the Company’s future performance’ {H/301/2}.”

¶ [141] It matters not that the term “ambitious but achievable” may well have been suggested to Mr Halder by an attorney in the process of preparing his written evidence for trial. The question is whether that term accurately reflected the way in which the estimates had been prepared. I accept Mr Halder’s characterisation of the estimates as being “ambitious but achievable” as truthful, and that this was not only the way in which he prepared his own forecasts but also the way in which he believed management projections were typically prepared80. Clearly during the Transaction process Mr Halder did not seemingly mention the limitations of the estimates; that they were not risk-weighted and that he always prepared “stretching forecasts”81. Nor did he, perhaps, mention that short-term estimates were likely to be more accurate than long-term estimates. There is no suggestion that he was asked about any such limitations by the Special Committee and omitted to mention them. In supporting the loan-raising efforts of the Buy Side which the Sell Side (Premier) would benefit from, it seems obvious that his “brief” was to “talk up” up future prospects, not to talk them down. In preparing his evidence for the present trial, his “brief” was clearly to focus on the limitations inherent in the estimates rather than strengths.

¶ [142] In my judgment there was no material inconsistency between Mr Halder’s pre-Merger Agreement position and his evidence at trial. The April and August 2017 Projections were both “current best good faith estimates of the Company’s future performance” (as he told the Special Committee) and “ambitious but achievable” as he told this Court. As I noted above, the Vice-Chancellor in In re Petsmart, Inc 2017 WL 2303599, upon which Mr Adkin QC relied in his opening oral submissions, adopted the following test for reliability in relation to management forecasts (at paragraph 32): 79 Paragraph 35 80 Transcript Day 2 page 226 line 10-page 228 line 6. 81 Transcript Day 2 page 50 lines 19-21. "The first key to a reliable DCF analysis is the availability of reliable projections of future expected cash flows, preferably derived from contemporaneous management projections prepared in the ordinary course of business. As this court has determined time and again, if the 'data inputs used in the model are not reliable,' then the results of the analysis likewise will lack reliability. And, as the experts in this case both agree, to be reliable, management's projections should reflect the 'expected cash flows' of the company, not merely results that are 'hoped for.'" [Emphasis added]

¶ [143] Professor Fischel cited an academic text in support of his opinion "a DCF analysis requires the use of expected cash flows, which reflect the probability-weighted average of all possible outcomes". On this basis, he contended that even with the adjustments proposed in Appendix D to his Report, the August 2017 Projections were "not well-suited for use in a DCF analysis"82. Such a strict reliability test was not supported by any judicial authority. Professor Gompers contended for a more flexible reliability test without seeing the need to cite any authority. As mentioned above, the Company argued in Lord Grabiner's opening oral submissions that Dell, Inc.-v-Magnetar Global Event Driven Master Fund Limited 177A.3d 1 (2017) (Supreme Court of Delaware) reflected a general move away from traditional reliance on DCF valuations in Delaware. Dell can be read as sounding a warning about relying wholly on a DCF analysis to produce an improbable appraisal result: the trial judge valued the company at $7 billion more than the transaction price and gave no weight at all to strong market price evidence (based on a large public float and a robust sales process). However, Dell does not support any new approach to determining when management forecasts are sufficiently reliable to be used to construct a credible DCF model. The management projections in Dell were held to be "obviously unreliable" in circumstances quite different to those in the present case: "The record simply does not support the Court of Chancery's favoring of management's optimism over the public analysts' and investors' skepticism—especially in the face of management's track record of missing its own projections. (Even Mr. Dell doubted his management team's forecasting abilities and conceded at trial, "We're not very good at forecasting.")... Given that we have concluded that the trial court's key reasons for disregarding the market data were erroneous, and given the obvious lack of credibility of the petitioners' DCF model—as well as legitimate questions about the reliability of the projections upon which all of the various DCF analyses are based—these 82 Fischel Report, paragraph 68. factors suggest strong reliance upon the deal price and far less weight, if any, on the DCF analyses."83

¶ [144] Mr Adkin QC in his closing oral arguments aptly relied upon the following passage in the Delaware case of Open MRI Radiology v Kessler, Court of Chancery for the State of Delaware, C.A. 275-N (Strine, Vice Chancellor): "The most important input necessary for performing a proper DCF is a projection of the subject company's cash flows. Without a reliable estimate of cash flows, a DCF analysis is simply a guess. Fortunately, Delaware Radiology was not engaged in a complex business and the operations of Delaware I and II provided a foundation for making some reasonable estimates of future performance. Of even greater utility, Carr of Tri-State made detailed projections of the performance of Delaware I, III, and IV for non-litigation use..."84 The projections for Delaware I were used in contemplation of securing financing from the Wilmington Savings Fund Society when Delaware I was being expanded and adding a second MRI magnet. The pro forma projections present the anticipated performance for Delaware I, including scan volume projections, for the two years from October 2003 through September 2005. These were provided to the Kessler Group in October 2003 when the Broder Group was requesting personal guarantees for Delaware I's expansion. The projections for Delaware III also were disclosed to the Kessler Group in October 2003, and these projections were used to secure financing for the equipment to be used at the Center. The projections for Delaware III, then, were prepared by October 2003. Similarly, Carr prepared two-year projections for Delaware IV to secure financing for that Center's equipment, which was provided by PNC Bank on March 9, 2004. Traditionally, this court has given great weight to projections of this kind because they usually reflect the best judgment of management, unbiased by litigation incentives. That is especially so when management provides estimates to a financing source and is expected by that source (and sometimes by positive law) to provide a reasonable best estimate of future results. Therefore, we have regarded with rightful suspicion attempts by parties who produced such projections to later disclaim their reliability, when that denial serves their litigation objective."

¶ [145] In the present case, the Management's Projections were represented to the Special Committee as "best estimates" and provided to Baring for deployment in finance- 83 At pages 44; 64. 84 At pages 64-65. raising efforts. They were used by Houlihan Lokey to prepare a DCF model which was relied upon together with other indicia as the basis of its fairness opinion. Any suggestion that the April and/or August 2017 Projections are entirely unsuitable for conducting a DCF analysis must be roundly rejected. Finding that forecasts are sufficiently reliable in general terms to be deployed in the context of producing a DCF model does not discharge the Court’s duty to critically assess, with the help of expert and factual witnesses, how reliable particular elements of the estimates are in the circumstances of the particular case. It is inherent in the nature of all estimates, as the Proxy Statement cautioned, that they are subject to various contingencies and risks and that they do not purport to represent “actual” future performance. This conclusion finds support in the previous decisions of this Court.

¶ [146] In Re Integra [2016 (1) CILR 192], Jones J was required to choose between a market-based valuation approach contended for by the company’s expert or a (75%) DCF valuation which was contended for by the dissenters’ expert. The company’s management projections had been prepared against the background of the pending merger and were used as the basis for the fairness opinion. The company’s expert contended that “management and the analysts have been repeatedly over-optimistic about the company’s anticipated performance”. Jones J, after noting this criticism, proceeded to tacitly accept without elaboration that the projections met the minimum requirements for reliability, subject of course to further scrutiny: “54…The fact that a high proportion of the value reflected in the experts’ DCF calculations is derived from the terminal value calculation means that the resulting value is very sensitive to small assumption changes. For these reasons it is particularly important that the cash flow projections/models are subjected to an in-depth review and analysis.”

¶ [147] In Shanda Games, for instance, both experts carried out a DCF valuation and Segal J resolved disputes about the extent to which the relevant management projections were entirely reliable in their original form or required adjustment in light of criticisms by the dissenters’ expert. It was common ground that the projections could be used to some extent. The “revenue intensity” column was a particular aspect of the management projections the reliability of which was disputed. Segal J resolved the dispute in the following way: “113. It seems to me that on balance the evidence suggests there was an error in the model which needed to be corrected. I also consider that in circumstances where Shanda had been given an opportunity to provide such an answer resolve the apparent errors and inconsistencies and demonstrate that their model and projections were reasonable) but failed to do so, I can and should infer and conclude against Shanda that there were errors and that the forecasts in this respect are unreliable. Furthermore, it seems to me to be right that, having concluded that there is an error which needs correcting, in the absence of any alternative corrections from Professor Jarrell (or evidence from Shanda), I should accept Mr Inglis’ evidence and corrections. His methodology seems to be reasonable and realistic and, in so far as he has adopted a correction that treats revenue intensity as being in need of an upward adjustment (rather than royalties and fees as being in need of a downward adjustment) this seems to be fair …”

¶ [148] The approach Segal J adopted to resolving disputes between the experts as to how future cash flows should be estimated was only partially subject to appeal and was ultimately approved by the Court of Appeal. These judgments illustrate that a DCF valuation can be relied upon by this Court even if the reliability of management projections are disputed in significant respects. Martin JA in In re Shanda Games [2018(1) CILR 352] described the disputes in the following way (at page 395): “64 In the present case, the judge had to resolve disputes between the experts relating to whether cash flows should be estimated in two stages or three; as to whether cash flows in the first period should be based on the company’s own management projections or on figures derived from the performance of similar companies; as to how certain elements of the company’s business, in particular the likely performance of one of its internet games, should be reflected in the cash flow estimates; and as to how the discount rate should be ascertained. Only three of his conclusions on valuation methodology are now disputed, all of them by the dissenting shareholders. The first two concern the judge’s assessment of beta (“the beta point”) and the small stock risk premium (“the SSRP point”) in the ascertainment of the discount rate; the third of them arises out of his decision to adopt a three-stage approach to the assessment of cash flows (“the transitional period point”).”

¶ [149] In Re Qunar Cayman Islands Limited, FSD No 76 of 2017, Judgment dated May 13, 2019, Parker J placed more reliance on the adjusted management projections in a section 238 appraisal case where both experts were seemingly agreed on the general reliability of the projections. Parker J explained his approach as follows: "Management Projections 177. As the basic premise underlying the DCF methodology is that the value of the company is equal to the value of the projected future cash flows (discounted to the present value at the opportunity cost of capital), the first part of the calculation involves estimating the values of future cash flows for a discrete period based on contemporaneous management projections. Both experts agree that the Management Projections are the most useful starting point when determining the value of the Company using the DCF method and so it is necessary to consider the important matters between them on this issue. 178. Since they were produced in August 2016 and the Valuation date is some six months later on 24 February 2017, the experts agree that the Management Projections should be updated to reflect the Company's actual performance. Actual results for 2016 and the 2017 budget had been produced by then and the results for Q1 2017 were beginning to be known. The experts worked from the updated projections. Approach 179. I accept the Dissenters' argument that the court should not defer to projections of the future performance of the Company simply because they have been made by those within the business at the relevant time. Where there are legitimate concerns that projections are unreliable, the court with the assistance of expert evidence and all relevant information available to it, is able to determine whether those projections are reasonable. 180. However, an important part of the analysis is what view the senior management of the Company came to and on what basis. They will ordinarily be in the best position to make reliable projections because it is they who have experience of running the actual business. In addition, country, sector and competitor knowledge will also be critical to the assumptions made in the forecasts. Ordinarily, absent a good reason to do so, one would not second guess the Management Projections. They are, after all, subjected to external scrutiny by market analysts on an ongoing basis and one might reasonably expect some contemporaneous challenge or public comment if that had come to light. 181. If nevertheless it can be shown that they are obviously wrong, careless, or tainted by an improper purpose (biased), that is a different matter and the court would revise them. 182. It is important to bear in mind that they are being challenged in litigation, ex post facto, through expert evidence. Neither expert is experienced in the OTA market in China, nor the Company’s specific business or strategy at the relevant time. An expert coming to a different view in this context is not a sufficient reason to make adjustments to the projections. The Company was making projections from real time information and knowledge. Here the analysis is to a large extent second hand, made some time later and involves second guessing judgments that were made at the time. It seems to me that I would need persuasive evidence to find that the projections made were flawed and to substitute my own opinion (or that of the expert’s) for that of the management at the Company and therefore adjust them in the ways Mr Osborne and the Dissenters require."

¶ [150] This valuable judicial commentary does not suggest a fundamentally different approach to that adopted by Segal J in Shanda Games. But Parker J’s observations do add an important gloss. If management projections are generally reliable, the Court should be cautious about allowing expert witnesses to effectively substitute their uninformed judgments about the company’s future business prospects for the judgments of those with intimate inside business knowledge. The degree of caution, to my mind, will usually depend on whether the disputed adjustments are indeed primarily dependent on questions of insider business judgment, rather than matters which are really questions of common sense or expert valuation methodology. However, Re Qunar also provides helpful guidance as to how to approach the issue of whether or not management projections are suitable in general terms as a basis for a DCF valuation analysis. Earlier in his judgment, Parker J made the following critical assessment of the way in which the management projections had been prepared: "112. Mr Zhu, the Company's former CFO, was responsible for preparing the Management Projections. I found him to be an intelligent and straightforward witness. He had experience in accountancy and banking. I formed the view that he understood the Company's business in depth, from which he was able to prepare the projections and he defended them on a reasonable basis under cross examination. 113. I have reviewed the transcript of the Management Meeting which took place on 7 November 2017 and I am satisfied that the evidence he gave in court is consistent with the answers he gave to questions put to him at that meeting, as well as the responses to the various information and data requests from the experts. At trial he strongly and credibly refuted any suggestion that the Management Projections were prepared from the point of view of financial self-interest or a desire to assist the majority shareholder to keep the share price low to effect the merger."

¶ [151] The approach Parker J effectively adopted to the threshold question of the general reliability test was to find that management projections can potentially be used as the basis for a DCF valuation where they have been prepared in good faith by a competent management team which understands the business and is capable of making informed judgments about future performance. This appears to me to be consistent with the tacit approach in the earlier local cases and the Delaware appraisal cases to which I was referred. Mr Halder described the Company’s April and August 2017 Projections as “ambitious but achievable” forecasts. The Company expressly advised Houlihan Lokey, for the purposes of their advice to the Special Committee dated April 25, 2017, that they could assume that the Projections had been “reasonably prepared in good faith on bases reflecting the best currently available estimates and judgments of such management as to the future financial results and condition of the company”85. In my judgment they meet the legally recognised minimum standards of reliability for use as a basis for a potentially credible DCF model in all the circumstances of the present case. They were not prepared with the starry-eyed frame of mind described by Jane Austen in ‘Sense and Sensibility’: “to wish was to hope, to hope was to expect”. After all, Mr Halder was a Chartered Accountant.

¶ [152] The April 2017 Projections were developed over time out of annual budgets and prepared under Mr Halder’s oversight. Although it seems clear that both the November 2016 Projections and the April 2017 Projections were prepared with the proposed sale of Premier’s majority stake to some extent in view, there is no suggestion that Mr Halder was forced to prepare his projections in an overly optimistic way. Nor is there any suggestion that any significant market actors viewed the Projections as wholly unreliable. The August 2017 Projections which were substantially based on the earlier Projections were, I find, sufficiently reliable (subject perhaps to some adjustments) to form the basis of a credible DCF valuation.

¶ [153] But, it bears repeating, this does not exclude the need in the context of a judicial appraisal proceeding to test the accuracy of the estimates and make appropriate adjustments with a view to determining the most realistic forecast of the Company’s future performance. Accepting management forecasts as sufficiently reliable for use in a DCF valuation does not mean that the Court cannot limit the weight to be given to the resultant valuation depending on the degree of reliability the Court can properly place on the estimates in question. 85 Proxy Statement Annex E, page E-3. The need to make downward adjustments to reflect adjustments made to the China Bilingual 5 Year Forecast How the need to consider making adjustments to the August 2017 Projections arises

¶ [154] As already noted above, Ian Johnson testified that in May 2017 Mr Andrew Fitzmaurice and Mr Graeme Halder asked Ms Tang and Mr Johnson to prepare a 5 year forecast for the China Bilingual business. The forecast model was ‘completed’ on July 26, 2017. The result of this analysis was projections far lower than the previous China Bilingual projections which had been based solely on the performance of the NACIS school in Shanghai. Account had to be taken of the fact that fee increases required regulatory approval. The lower forecast projections were also because, inter alia, the proposed schools were to be set up in smaller, less wealthy and more remote cities, resulting in likely far lower full-time enrolment (“FTE”) targets being achieved. Mr Halder received this update before the August 2017 Projections were completed, but was so busy at the time that he failed to consider it and incorporate it into those projections. However, he conceded that he was aware “at a high level” that the numbers had to come down and made some partial downward adjustments in the August 2017 Projections86. The way in which Mr Halder explained the failure to give full effect to the China Bilingual 5 Year Forecast was perhaps the least convincing aspect of his testimony, albeit on a very peripheral matter.

¶ [155] It seemed more likely to me during his evidence that Mr Halder was somewhat uncomfortable with the notion of making significant downward adjustments to more optimistic projections he had recently been involved in ‘hawking’ in the lender presentations. On any view, if his professional approach was to prepare forecasts which had no risk weighting in them, it would be odd for one slice of the August 2017 Projections to have risk weighting while most of the forecasts did not. Another possible explanation for their omission is that the new model was incomplete and only preliminary in character. Whether or not the China Bilingual downward adjustments were omitted in full-blown form by accident or design is really by the way.

¶ [156] Not only did Mr Johnson contend for modifying the China Bilingual element in the April 2017 Projections. He also supported the Company’s case by testifying under cross-examination by Mr Adkin QC that the Parthenon Reports did not provide a solid basis for projecting future cash flows. This was because despite the general utility of those reports in terms of providing background research “they failed to take into account any of the difficulties and the barriers to entry.”87 The Dissenters contended that the true value of this project was being understated. Although it is entirely understandable that the Dissenters should cry “foul”, I can find no basis for doubting the straightforward, coherent and clear evidence of Mr Johnson on this issue. Before a 86 Transcript Day 2 page 204 line 24-page 205 line 21. 87 Transcript Day 3 page 20 lines 2-14. detailed analysis of the China Bilingual project was carried out, reliance was placed on the past performance of one successful school. This aspect of the April Projections gave a rose-tinted view of the prospects; so did the Parthenon Reports which were deployed by Baring when making pitches for lending. But there was no deception, because it was clear on the face of the relevant document that no specific projections had been made in relation to projected future schools.

¶ [157] Would it be naïve to believe that the timing of Mr Johnson’s work, after the Merger Agreement and before the EGM which would likely trigger dissenter litigation, was entirely neutral in its effect on the contents and of the China Bilingual 5 Year Projections? While I accept the core integrity of the analysis that was carried out, I do not ignore the fact that the work carried out by Mr Johnson and his colleague Ms Tang (who was said to possess valuable local knowledge) appeared designed to identify downside risks rather than to carefully evaluate the appropriate weight to be given to both upside and downside risks. Professor Gompers objected to these new Projections on the grounds that they appeared to be “too conservative”88. To the extent that Mr Johnson conceded that the China Bilingual Model he emailed to Mr Halder on July 2, 2017 contained numbers which were “a work in progress”, and the document was acknowledged by Mr Fitzmaurice as a “a useful start”, caution is required in evaluating what adjustments should indeed be made89. Further work was done on this draft after Mr Halder commented in an email dated July 21, 2017 on what he described as “a good start”. The model (“Bilingual Schools-Rolling Forecast Model-26 July 2017”) incorporating these comments was sent back to Mr Halder (copied to Mr Fitzmaurice) by Mr Johnson by email dated August 1, 2017.

¶ [158] Mr Halder’s Second Affidavit gives a fulsome explanation as to how it is (in a busy time leading up to the EGM) that he failed to appreciate that the new model justified a materially lower forecast for that business segment than was suggested by the August 2017 Projections. He fairly points out the latter document was only intended to be for internal management use. Mr Halder avers that “these projections could and should have been subject to adjustment for Mr Johnson’s China Bilingual projections and risk-weighting (please see paragraph 24 below)” (paragraph 23). He then explains (in paragraph 24) that his projections are not usually risk-weighted: “…This arises in part from my desire to stretch the business, in part from the difficulty in modelling the risks and in part from the fact that modelling risk weighting would not have materially improved our day-to-day decision making because the business was sufficiently strong to withstand individual schools materially underperforming...and as a business we had already 88 Supplemental Report, paragraph 294-300. 89 Transcript Day 3 page 6 line 12-page 8 line 4; page 10 line 16-page 11 line 1. decided to accept the large risks associated with doing business in many of the markets we were in, particularly China."

¶ [159] The assertion in a single sentence that "these projections could and should have been subject to adjustment for Mr Johnson’s China Bilingual projections and risk-weighting" is somewhat lukewarm support for the Company’s position on this issue. No explanation is offered as to why risk-weighting should, contrary to his historic approach, have been applied exclusively to this aspect of the Company’s Projections. And Mr Halder in explaining his own contrary approach to projections casts doubts on both (a) the viability of undertaking a risk-weighting exercise at all, and (b) the utility of such an exercise as a practical management tool. It is also unsatisfactory that as the Company’s Chief Financial Officer at material times who was ordinarily in charge of preparing Projections, it was unclear to me when, if at all (prior to his Second Affidavit sworn on April 10, 2019 after the termination of his employment), Mr Halder himself actually blessed the China Bilingual Model as an update to the August 2017 Projections.

¶ [160] Professor Fischel appeared to regard it as obvious that the August 2017 Projections should be adjusted "to reflect management’s latest views on the prospects for its China Bilingual initiative"90. Professor Gompers countered that July 2017 China Bilingual modifications “are incomplete and do not appear to be a measure of expected cash flows, which makes them unreliable to be used in a DCF valuation”91. The Company succeeded in demonstrating (principally through Mr Johnson’s evidence at trial) that the July 2017 China Bilingual analysis was obviously more sophisticated (or detailed) than the previous analysis, so I am unable to accept Professor Gompers’ broad-brush dismissal of the work described by Mr Johnson. On the other hand the Dissenters succeeded in raising doubts about what to make of the somewhat curious circumstances in which it is contended that the August 2017 Projections should be adjusted on the basis of the first set of risk-weighted projections conveniently prepared on the eve of the EGM with a whiff of dissenter litigation in the air. Accordingly, in light of the importance of this question to his valuation, Professor Gompers’ more detailed critique of the extent to which this new material should be used to make the adjustments to the August 2017 Projections which were applied by Professor Fischel requires more careful assessment. 90 Expert Report, paragraph 67. 91 Supplemental Expert Report, paragraph 289. China Bilingual 5 Year Forecast: was it incomplete?

¶ [161] Professor Gompers in his Supplemental Report92 opined that the Model finalised on July 26, 2017 was incomplete because a Summary page which had projections for a total of 8 schools leaving blank placeholder rows for another 9 schools. The April 2017 Projections contemplated 20 school openings through FY2023 while the August 2017 Projections contemplated 22 openings during the same period. Responses to information requests during the trial preparation period indicated that the Company contemplated 3-5 new schools each year from FY2020 through FY2024. Professor Fischel countered in his Supplemental Report the responses to management questions were in fact framed to reflect current views of openings, qualified by statements about the uncertainties and that accordingly the July Model’s forecasts could not be said to be incomplete. Professor Gompers accepted under cross-examination that the post-Valuation Date views of Management could not be used in a DCF calculation, but explained that he was simply relying on the responses to requests for information to explain why it seemed more reasonable to rely on the similar school openings numbers in the August 2017 Projections, which were far higher than the numbers in the July 26 2017 Model93.

¶ [162] Despite the somewhat equivocal way in which Mr Halder asserted in evidence that the August 2017 Projections should be adjusted to include the China Bilingual 5 Year Forecast prepared by Mr Johnson and finalised on or about July 26, 2017, I find on balance that the latter document does reflect Management’s revised and risk-weighted views as to expected future performance. Those views cannot be dismissed altogether on the grounds that it was incomplete. Moreover, as Professor Fischel rightly pointed out94, although the July 26, 2017 Model’s ‘China Bilingual Schools Forecast 5 Year Summary’ page reduced the projected number of schools, it still projected a substantial increase in capacity. That suggests a more balanced and considered assessment of future performance than is implied by looking solely at the number of schools. Moreover, it was clear from the terms of the response to the information request put to Professor Fischel by Mr Adkin QC in cross-examination, that the interviewed Company representative was only expressing hopes as opposed to expectations95. This prompted Professor Fischel to observe: “there is a difference between hopes and expects”.

¶ [163] In the end, the reliability of the competing valuation approaches to the China Bilingual project is best confirmed by cross-checking them through a broad-brush commercial judgment of the relative plausibility of the projected outcomes. 92 Paragraph 291. 93 Transcript Day 10 page 5 line 15-page 6 line 14. 94 Supplemental Report, paragraph 78. 95 Transcript Day 8 page 3 line 4-page 5 line 11. China Bilingual 5 Year Forecast: how much value?

¶ [164] The July 26, 2017 Model contemplates 6 new schools over the relevant period. Professor Gompers projects 22 based on the August 2017 Projections, and Professor Fischel projects 9 new schools (adding an additional 3 to the Company’s adjusted Projections). Accepting that this approach is perhaps an overly conservative one, relative to the approach the Company generally adopted to preparing its forecasts, I prefer Professor Fischel’s general approach on the grounds that there is no credible basis for continuing to rely on the August 2017 Projections as regards the expected number of schools. On balance I found Professor Fischel’s approach to this issue, while conservative, to be more objective and commercially realistic than Professor Gompers’ comparatively inflexible analysis.

¶ [165] The commercial result of Professor Gompers’ approach to valuing the China Bilingual business, assigning US$3 billion in value, was simply incredible. I accept the following assessment of the evidence on this point in the Company’s Written Closing Submissions: “141. The fact that the August 2017 Projections used by Professor Gompers in his DCF valuation were far too optimistic as regards China Bilingual was starkly revealed on Day 9. Professor Gompers was seemingly unaware {Day10/20:1} - {Day10/21:2} that the value attributed to China Bilingual in his valuation was approximately $3 billion ($29 per share); although he did not express any surprise at that figure {Day10/29:1-6}, he was wholly unable to explain how a business with a single operating school could be worth anything like that amount or to reconcile that figure to the $1 billion market value of China Maple Leaf [Exhibit XIII-2 at {E/21}], one of Nord’s closest competitors (per Professor Gompers Exhibit XIII-2 at {E/21/1} (he agreed that was his description on Exhibit {Day10/34:9-17} although he backtracked on Day 12) and the owner and operator of 71 bilingual schools in China {HSD/1/147}. Subsequent attempts to compare an operator of schools, many of them not-for-profit schools, to Apple and the launch of a new iPhone did not enhance his credibility {Day10/88:1-18}. See also Qunar where the Dissenters’ expert had valued the company at more than Expedia [{AB/8/33} paragraph 175].”

¶ [166] The attribution of US$3 billion to China Bilingual by Professor Gompers was asserted by the Company’s counsel in opening and not contradicted by Professor Gompers under cross-examination. It does not appear on the face of his Reports. It may well be that the $3 billion figure is an extrapolation from Professor Gompers’ $1.454 billion “Present Value of Incremental China Bilingual”. The latter figure formed part of a global total equity value of US$8.1 billion (also presumably discounted to present value) which Professor Gompers used in his Expert Report to support a diluted share price of $76.1296. In any event I regard that global valuation, more than twice the Market Price and the Transaction Price, as inherently improbable, for reasons set out near the end of this Judgment.

¶ [167] That said, the unusual circumstances under which the China Bilingual 5 Year Forecast was prepared with the introduction for the first time of risk-weighting after the Merger Agreement had been consummated cannot be considered as having no impact on the reliability of the adjustments which Professor Fischel himself adopted. It is not necessary for me to decide whether the Company was engaging in sharp practice by trying to massage the estimation evidence in anticipation of the present litigation. It suffices to conclude that because the China Bilingual 5 Year Forecast was prepared on a more conservative basis than the August 2017 Projections, there is a risk that the former estimates are somewhat understated and a risk that the latter estimates are somewhat overstated. China Bilingual: summary

¶ [168] In summary, I approve Professor Fischel’s adjustments to the August 2017 Projections to take into account the China Bilingual 5 Year Projections. While it was unattractive for the Company to prepare for the first time risk-adjusted estimates for an important part of its business at the time when it did, I accept that overall the adjustments increased rather than decreased the reliability of the estimates. This is essentially because the revised estimates were based on far more relevant data. DCF Methodology: overview

¶ [169] In his Expert Report, Professor Fischel opined as follows: “62. A DCF analysis also requires an estimate of the ‘terminal value’...The terminal value in a DCF is typically calculated using the “Gordon Growth Formula”, under which the value of the company at the end of the explicit forecast period is equal to \( \text{FCF}_{t+1/r-g}/(r-g) \), where 'FCF_{t+1/r-g}' is the expected free cash flow in the first year after the end of the explicit forecast period, 'r' is the discount rate, and 'g' is the assumed annual growth rate of free-cash flows 96 Expert Report, Exhibit XI-3A. into perpetuity (which is commonly referred to as the 'perpetuity growth rate' or 'PGR'). 63. Once the terminal value is estimated, it is discounted to present value and added to the estimated present value of the free cash flows in the explicit forecast period to estimate the value of the company's operating business. Then, the company’s total enterprise value is estimated by adding the estimated value of cash and cash equivalents and non-operating assets (such as unconsolidated subsidiaries, and other equity investments) to the estimated value of the company’s operations. The aggregate equity value of the company is estimated by subtracting the estimated value of debt and other non-equity claims from the estimated total enterprise value of the company. The value per share is estimated by dividing the aggregate equity value by the number of fully diluted shares outstanding."

¶ [170] Essentially the same methodology is described by Professor Gompers in paragraph 219 of his Expert Report. As indicated above, the Experts were agreed on the key elements of a DCF model. The Bach Bank Model and other contemporaneous views on value

¶ [171] Professor Gompers carried out two DCF analyses, one based on the August 31, 2017 Projections and the other based on the Bach Bank Model prepared by Baring, not the Company’s Management. In my judgment the Bank Model clearly cannot be used as the basis for a DCF valuation because, I find as a fact, despite the fact that it was prepared with input from the Company’s Management, that it was not prepared in the ordinary course of the Company’s business as a tool for predicting its future performance. I accept the evidence of Mr Halder in this regard. It seems obvious from all the relevant evidence that the Bank Model was used by Baring to raise financing in connection with the Merger Transaction and/or for monitoring the investment in the post-Merger period.

¶ [172] In section B of his Supplemental Report, Professor Fischel opines that the Bank Model projected how the Company would perform post-Merger. In addition, it was not a model prepared by Management even though it drew on information provided by Management. Professor Gompers in his Reports advances no convincing rationale for using the Bank Model as a basis for a DCF analysis.

¶ [173] Professor Gompers, and the Dissenters, placed considerable reliance on the views of Baring and various individuals connected with the Company to the general effect that the Shares were undervalued by the market and/or had an intrinsic value far higher than the Transaction Price. This evidence was used to support the credibility of the far higher DCF valuation contended for by Professor Gompers rather than the more modest range proposed by Professor Fischel. I do not find any credible support from such sources for a valuation at the remarkable level contended for by Professor Gompers. However I am bound to accept the following far more modest and factually supported argument advanced in the Dissenters’ Written Closing Submissions, which does no more than to support in general terms the need to have regard to an appropriate DCF analysis: “288. A telling way of testing whether DF’s refusal to give any weight to a DCF analysis in this case is a reasonable and realistic one is to look at what the economic actors and other market participants did. When this is done, the picture is stark. All of the economic actors operating in the real world used a DCF analysis in seeking to establish the value of the Company and its shares, and every one of those DCF analyses showed the stock to be undervalued by the market. Conversely, not a single one of them adopted the approach used by DF and tested the efficiency of the market to arrive at the conclusion that the market price should be relied on.”

¶ [174] I accept that the credibility of the Market Price, which was primarily relied upon by Professor Fischel as the best evidence of the fair value of the Shares, is materially undermined by the extent to which various players had recourse to DCF models. However I do not find that any of these comparatively informal analyses provides any meaningful support for Professor Gompers’ ultimate US$76.51 DCF valuation outcome. Terminal Growth Rate “TGR”/Perpetuity Growth Rate (“PGR”)

¶ [175] Professor Fischel explained his approach to terminal value in his Expert Report as follows: “78. If the Company were to make replacement capital expenditures that sufficed to maintain the productive capacity of its assets but made no new capital investment, then the company’s PGR would be equal to the long-run expected inflation rate, which was 1.81% as of August 1, 2017. In order to obtain a larger PGR, the Company would have to reinvest a portion of the after-tax operating profits it earns back into the Company and earn a positive return on that reinvestment. Therefore, the assumed PGR must be consistent with assumptions made about the amount of, and return on, new investments made in the terminal period. 79. I understand that Nord Anglia education’s existing schools operate in competitive markets...Moreover, economic theory indicates that returns that exceed a company’s costs of capital will be competed away over time. Accordingly, our DCF analysis assumes that Nord Anglia Education’s rate of return on new investment capital (‘RONIC’) will be equal to its WACC during the terminal period. This assumption is consistent with the concept of a steady state. Under those circumstances, the amount of new capital investments would affect the expected PGR but would not affect the terminal value. Therefore, the results of our DCF analysis do not depend on the assumed amount of new capital investment in the terminal period.” [Emphasis added]

¶ [176] In Appendix D, Professor Fischel explains in further detail how the projections are extended to reach steady-state. He adopts a 1.81% long-term growth rate (TGR or perpetuity growth rate) based on the rate projected in the Company’s Adjusted August 2017 Projections for 2027, the last year of the main forecast period.

¶ [177] Professor Gompers in his Expert Report opined as follows: “313. Steady state cash flows are cash flows generated after the explicit forecast period, once a company is in a steady-state growth period. It is generally assumed that, at some point, the company’s cash flows follow a predictable, steady state pattern. This steady-state cash flow is used to calculate the value of the company, which is assumed to exist in perpetuity. Nevertheless, even though infinite life is assumed for the company, cash flows that occur far out in the future receive progressively less and less weight (the weights become closer and closer to zero as the horizon increases) due to the nature of discounting for time value of money and risk using the discount rate (WACC). 314. Overall, I take a conservative approach (resulting in a lower valuation) to the calculation of the steady-state cash flow. My approach is conservative because I assume that Nord Anglia will stop growing its footprint in the terminal period...This means that any growth in cash flow during the terminal period comes only from increases in tuition. This is conservative because, as discussed previously, Nord Anglia had a proven track record of identifying and completing value-enhancing acquisitions, and the Company expected to continue those acquisitions going forward… 343. It is a common assumption in DCF valuation that the long-term growth rate for a company is expected to fall between the expected rate of inflation and the expected nominal GDP growth rate. Weighting the expected nominal GDP growth rates for 2022 in the countries in which Nord Anglia operated by the Company’s last year of projected revenue in the August 2017 Projections and Bank Model in those countries yielded a weighted average expected nominal growth rate of 6.68% and 6.21% using the August 2017 Projections and Bank Model respectively. The growth rates I use are significantly below the GDP growth rates.” [Emphasis added]

¶ [178] In his Supplemental Report, Professor Gompers further explained his approach as follows: “247...the growth rate I use is equal to 1.5x the weighted average (by revenue) inflation rate for Nord Anglia, based on the Company’s historical and targeted tuition fee increases at 1.5x to 2.x inflation. The August 2017 Projections and the Bank Model make different assumptions about the exact geographical mix of revenue for Nord Anglia in the future. Therefore, the weights associated with inflation in each country differ to the extent the geographical mix of revenue is different across these projections. As a result, a growth rate equal to 1.5x weighted average inflation rate will be different as well. This implies a long-term growth rate of 4.23% for the valuation using the August 2017 Projections and 3.99% for the valuation based on the Bank Model.”

¶ [179] With the Dissenters’ Expert propounding a long-term growth rate of 4.23%, more than twice the Company’s Expert’s 1.81%, Professor Fischel understandably strongly criticised the Gompers approach, invoking Professor Gompers’ own writings in support of his analytical cause. In his Supplemental Report, Professor Fischel opined as follows: “86...if the Company’s tuition revenues were expected to grow at 1.5 times the local currency inflation rates of the countries in which the Company operates in, as Professor Gompers assumes, then the U.S. dollar value of the Company’s tuition revenue would be expected to grow at 1.5 times the U.S. dollar inflation rate. As of August 1, 2017, the long-run expected U.S. dollar inflation rate was 1.81% 1.5 times that amount is 2.715%, not the 4.23% or 3.99% U.S. dollar growth rates that Professor Gompers assumes.” 87. Professor Gompers’ assumption that the Company’s long-term tuition growth rate would substantially exceed the expected inflation rate is also implausible...Moreover, in his recently published casebook, Professor Gompers states that ‘[w]e often assume...zero real growth in the future’. A zero real growth rate would imply a long-term nominal growth rate equal to the expected long-run inflation rate, or 1.81% in U.S. dollars, not the 4.23% or 3.99% growth rates that Professor Gompers assumes. 88. As Professor Gompers states in his recently published casebook, ‘[v]aluation[s] in which the terminal value represents a substantial majority of the value are likely to be very sensitive to small changes in growth rate assumptions’. That is precisely the case here, as terminal value represents approximately 94.1% of the Company’s enterprise value in Professor Gompers’ DCF analysis based on the August 2017 Projections and approximately 87.2% of the Company’s enterprise value based on the Bank Model....”

¶ [180] The differences between the Experts were summarised in the Dissenters’ Written Closing Submissions as follows: (a) whether the growth rate should be at or above inflation; (b) what inflation rate should be used; and (c) whether return on capital will equal the cost of capital in the terminal period.

¶ [181] Professor Gompers persuasively argued (based on facts that I considered to be uncontroversial) that it was reasonable to assume that the Company’s tuition fees would increase at 1.5 times the rate of inflation in the countries the Company did business in having regard to the profile of those countries, historic returns, the particular prospects of education as a vibrant (as opposed to dying industry) and taking into account that in reality the perpetuity period was only really taking into account 30-40 years. Under cross-examination by Mr Boulton QC, he admitted that the question of whether or not an assumption should be made that growth would be higher than inflation had significant financial consequences for the valuation question: “Q. So if we remove the 1.5 times multiplier and simply use the weighted average inflation rate, your 4.23 per cent would go down to 2.82 per cent. Is that right? A. Correct. Q. Let's insert that at B45, simply to see the sensitivity that arises from that assumption. 2.82 per cent, please, operator. $54.10. So something like $21 of value arises simply from an assumption that the company for ever more will be able to increase its tuition fees at 1.5 times the rate. I think that may be an underestimate."97

¶ [182] In my judgment Professor Fischel’s assumptions, which align with the economic theory which predicts a loss of competitive advantage over time, and take into account the sensitivity of the valuation to minor changes in growth rates in cases such as the present, are more reasonable overall. I reject Professor Gompers’ approach of applying a 1.5 multiplier to the inflation rate on the basis that I find that it is based on an optimistic view of the Company’s long-term fee-generating capacity with insufficient consideration being given to downside risks.

¶ [183] What should the terminal growth rate be? Professor Fischel in his Supplemental Report did not appear to me to challenge Professor Gompers’ view that the terminal growth rate should ordinarily be somewhere between the inflation rate and GDP. Mr Boulton QC also appeared to accept Professor Gompers’ position on the latter issue in cross-examination98. The main justification Professor Gompers advanced for going above the inflation rate was applying the 1.5 multiplier based on an assumed 1.5% fee increase through the terminal period. The main challenge put to the Dissenters’ Expert on this point, as already noted, was that his general approach was simply inherently unrealistic: "Q. Do you not find it worrying that half of your value turns on a debate about whether one should be using a US inflation rate or 1.5 times local inflation rates in a period that starts ten years from now? A. So does it concern me? It certainly -- I certainly think about it. It doesn't change my opinion. I think -- as I mentioned earlier, the cash flows that Professor Fischel and I get are sort of reasonably close. I think the important issues for this court to think about is what is that long run terminal growth rate, as well as the discount rate, which I'm sure we will have a discussion, but I agree with you that if you change the long run terminal growth rate to 1.81 per cent, which I think is absolutely wrong, it causes the value to fall by nearly half." 97 Transcript Day 10 page 123 line 25-page 124 line 3. 98 Transcript Day 9 page 105 lines 9-15.

¶ [184] The Dissenters in their Written Closing Submissions also relied on representations made by the Company to Houlihan Lokey about the terminal growth rate which were seemingly accepted by the advisers to the Special Committee: "432. Importantly, in the context of the take-private, the Company told Houlihan Lokey, as advisors to the Special Committee, that tuition in the steady state would grow at 1-2% above inflation. This representation was not simply directed at the immediate future, but at the terminal period..." 433. Houlihan Lokey relied on that message in the presentation they gave to the meeting of the Special Committee on 24 April 2017 and then to the Company’s Board at their meeting on 25 April 2017, a presentation which came to be filed with the Proxy. Houlihan Lokey included a sensitivity table showing equity values at different discount rates and TGRs. The alternative TGRs in the steady state were 3.50%, 3.75% and 4.00%. Footnote 4 on that page makes clear that 'revenue growth [was] based on information provided by Company management'."

¶ [185] This submission does not support Professor Gompers’ use of a growth rate above the rate of inflation because, whatever growth rate Houlihan Lokey used for their DCF analysis, it produced a range of values far closer to Professor Fischel’s valuation results than to Professor Gompers’. I accordingly approve the use of the [US$] long-term inflation rate of 1.81% as the terminal growth rate because on a balance of probabilities it appears to me to be based on the most reasonable assumptions it is possible in all the circumstances to fairly make.

¶ [186] The Dissenters finally complain that Professor Fischel’s assumption that the Company will make new investment in the terminal period but that any return on investment will not exceed the cost of capital to be “extraordinary”. As Professor Gompers assumes no growth in the terminal period at all, the practical valuation result appears to be the same, albeit via different analytical routes. In these circumstances I see no need to decide to prefer either approach, being firmly of the view that no justification exists for a terminal growth rate above the long-term inflation rate.

¶ [187] In summary, I approve the terminal growth rate of 1.81% as proposed by Professor Fischel in his Supplemental Report]. Cost of equity: Capital Asset Pricing Model ("CAPM")

¶ [188] Both Experts used the Capital Asset Pricing Model ("CAPM") to calculate the Company’s cost of equity. The formula used is the risk-free rate of interest + the beta x the equity risk premium. The formula is mathematically expressed as *cost of equity = Rf + β × ERP*. The main dispute at trial was the appropriate *beta*, the appropriate risk-free rate of interest and the equity risk premium being agreed. Cost of equity: the appropriate discount rate and “beta” Overview

¶ [189] The Company submitted that using Professor Fischel’s *beta* number "in Professor Gompers’s August 2017 DCF calculation would reduce his share price from $75.21 to $51.69"99. So I cannot ignore the fact that what appears to be a narrow technical dispute is nonetheless in commercial terms a “big ticket item”. Identifying an appropriate beta is the first step in calculating the actual discount rate which is designed to determine the present value of the Shares based on an assessment of the likely future cash flows.

¶ [190] Professor Fischel in his Expert Report described the function of the discount rate in a DCF analysis as follows: "61. *A DCF analysis is a method of estimating the value of an asset (i.e., what a willing buyer would pay a willing seller for that asset). In a typical DCF analysis, the expected stream of free cash flow available to investors in the company (i.e., equity holders, debt holders and any other nonequity investors) for an explicit forecast period (e.g., five years) is discounted to present value using an appropriate discount rate in the first stage. The discount rate in a DCF analysis is calculated using a company’s weighted average costs of capital ('WACC'), which represents the opportunity costs of capital of investors in the company. As suggested by its name, the WACC is a weighted average of the company’s cost of equity and its cost of debt, where the weights are based on the share of each type of capital in the company’s capital structure.*"

¶ [191] Professor Gompers in his Expert Report opined as follows: "215. *One of the central tenets of modern finance is that the value of an asset, such as a share in a company, is equal to the discounted value of an asset’s expected after-tax cash flows. A valuation textbook notes that DCF valuation is 'the foundation on which all other valuation approaches are built.' In the 99 Company’s Written Closing Submissions, paragraph 131.4. present circumstances, the objective is the calculation of the fair value of the Dissenters’ shares in Nord Anglia. Because shares are claims on the value of an underlying company (equity being a residual claim after debt holders have been paid), the starting point in valuing shares is the overall value of the company as a standalone entity, which is then adjusted to account for outstanding debt. The residual value is then divided by the number of shares outstanding to arrive at the per share valuation."

¶ [192] It is helpful to bear these uncontroversial statements of basic principles in mind when considering the controversial elements of their application to the facts of the present case. The dispute on the calculation of the discount rate ultimately involves a choice between Professor Gompers’ lower 1.1% figure which resulted in a WACC of 8.23% (without a size premium) and Professor Fischel’s 1.3% rate, which produced a WACC ranging from 8.98% (without a country risk premium or a size premium) to 10.90% (with both such premiums taken into account). Accordingly, the most significant threshold dispute in relation to the DCF analysis is how beta should be calculated. The two main disputes were: (a) what weekly data should be used for computing the beta; and (b) whether the resultant beta figure should be adjusted downwards using what is known as a ‘Blume adjustment’. The correct approach to weekly data

¶ [193] In the Company’s Written Closing Submissions, its position on the weekly data issue was summarised as follows: “146 The textbooks show that beta is commonly estimated using either two years of weekly returns or five years of monthly returns (Damodaran on Investment Valuation at {G/85/84}). There is a trade-off between obtaining more data points and going further back before the Valuation Date, in which case there it is more likely that the risk profile of the company will have changed. 147 In this case, Professor Fischel has estimated Nord’s beta using weekly returns (Friday to Friday) for the two year period before the Announcement Date, regressed against the S&P 500: 147.1 Nord had only been public for three years and therefore it would not have been possible to use five years of monthly data; 147.2 Professor Fischel’s approach is wholly consistent with the default approach taken by Bloomberg and many other data providers (whose business is to provide this sort of data to their clients); 147.3 Professor Fischel’s approach is therefore entirely normal and it cannot be suggested (and was not suggested) that he was in any way influenced by what the raw data shows. 148 Professor Gompers, in contrast, has adopted an approach of his own devising which he has not previously used and which is not the subject of any academic literature {Day 10/164} and {Day 10/166}. (The Dissenters sought for the first time in re-examination to suggest that Mr Osborne had applied the same method in Qunar, but a “five day trailing average” is not the same approach as adopted by Professor Gompers. Further it is a remarkable coincidence that these two “innovative” approaches were used by independent experts on behalf of the same clients in s.238 proceedings, and the court is respectfully asked to be sceptical about their provenance).”

¶ [194] In the Dissenters’ Written Closing Submissions, the following contrary case was advanced: “339. It follows that the principal differences between the Experts relate to (i) the length of the estimation period and (ii) how to use the weekly data. PG uses as much of the available data as possible in order to reduce estimation error in his regression analysis. He then cross-checked his result against other methods of arriving at beta. By contrast, DF appears to have selected a series of simplified parameters, each of which serve to increase the beta; he did not cross-check his beta against any other results. 340. PG’s approach should be adopted as the empirically rigorous approach, which has support in the academic literature. Despite DF’s suggestions to the contrary, DF’s approach is not recommended in any of the material before the Court. The Experts’ approaches to beta 341. PG arrived at his beta of 1.04 in a scientific way, as follows: 341.1. He chose the longest possible period for the comparison between the volatility of returns of the Company’s stock, as compared to a stock index, from 22 September 2014 to 24 April 2017 (a period of 2.6 years). This is the longest period of time because it includes all available weekly return data for the Company up until the last day prior to the Merger Announcement, with the exception of the first 180 days after the Company’s IPO which were excluded due to potential anomalous trading behaviour during this period which has been documented in finance academic literature. 341.2. As an index, he chose his return data from the broad CRSP value-weighted NYSE/AMEX/Nasdaq/ARCA return index. However, he also cross-checked this against the S&P 500 returns, which was the index used by DF. In any event, the Experts agree that in this case their different choice of index makes no material difference. 341.3. PG observed the weekly returns for each day of the week (Monday-to-Monday, Tuesday-to-Tuesday, Wednesday-to-Wednesday, Thursday-to-Thursday, and Friday-to-Friday) against the index returns, and carried out five separate regression studies resulting in 5 separate results: for these the lowest was 0.97 (with two others at 0.98 and 0.99), and the highest was 1.24. He averaged the 5 (see rows 41 to 45 in column c of the summary sheet tab/spreadsheet in exhibit A-1 {E/23}) to reduce estimation error. This is a case of using more data when available, but without the risk attendant on taking data from periods in the more remote past. 341.4. He applied a Blume adjustment because computing the cost of equity is forward-looking, and the derivation from past returns is backwards looking. The Blume adjustment (which can be compared with the smoothing adjustment made by Bloomberg) produced an adjusted beta of 1.04 (equivalent to the relevered beta of 1.01). 341.5. PG compared his beta of 1.01 against the results that are produced by other recognized methodologies used to arrive at beta198. In particular, he crosschecked and found his beta consistent with (1) the beta produced by Barra Inc (0.88), (2) betas applied by analysts covering Nord in 2017 (typically around 1.00), and (3) three US-traded (that is NYSE-listed) companies in the same education sector as the Company, viz. Bright Horizons, New Oriental, and TAL Education. He also used as a comparator the Fama-French Three Factor Model using two different ways of using that model to calculate the cost of equity – both of which produced a lower cost of equity (9.35% and 8.67% respectively) than his use of the CAPM with a historical adjusted beta201. 342. DF estimated the Company’s beta to be 1.30. He did so simply by taking returns from Friday-to-Friday, with weekly intervals, and using an estimation period of 2 years. He did not take into account the 2-year weekly results of any other day of the week. He did not take into account the data available over the longer period used by PG, viz. 2.6 years. He performed no cross-check, whether against other methodologies or contemporaneous estimates of beta, when arriving at his beta. He rejected his own 3 year study, used when estimating volatility in the context of his event study, to come up with the highest beta possible. DF’s beta should be rejected 343. DF’s beta should be rejected for the following reasons. First, he has selected Friday-to-Friday, which yields the highest beta, for no principled reason and without advancing any academic text that recommends such an approach. Second, he has chosen to limit the estimation period to 2 years, when 2.6 years of data is available and can and should be used because it reduces estimation error. Third, he has not cross-checked his beta against betas derived from any other methodologies, and a cross-check reveals that DF’s beta is an outlier. 345. By choosing only to use Friday-to-Friday data, DF has arbitrarily excluded data which could be taken into account in arriving at beta, and which should be taken into account in order to reduce estimation error. As PG says, "There is no basis in economic theory to prefer any one seven-day period vs. another, because all the estimates are estimates of the same quantity – beta. Therefore, the appropriate approach would be to minimize estimation error, which is what I have done."

¶ [195] I found this issue difficult to initially evaluate as, bearing in mind the broad consensus between the Experts on many technical matters, it seemed somewhat surprising that they should each contend that the other was adopting an approach wholly unsupported by academic literature. The first question seems to me to be whether it matters which seven day period one uses. Professor Fischel, under cross-examination by Mr Bompas QC, initially asserted that there was no logical basis for different results over different seven day periods and so any differences should be disregarded. He stated that the Bloomberg approach of Friday to Friday was standard and there was no support for Professor Gompers’ averaging approach100. When it was suggested that in fact using different weekly periods yielded different results (and his Friday to Friday period happened to generate a higher beta in this case), Professor Fischel insisted that such differences should be disregarded rather than taken into account, because the differences made no sense101.

¶ [196] Professor Gompers under cross-examination by Mr Boulton QC admitted that his using different days of the week and averaging was not a widely-used approach to calculating beta. He also admitted that he was unaware of any academic support for what was effectively a novel approach: "Q. Let's see how you describe what you've done and I'm going to come on to this averaging of five sets of weekly returns. Let's go to {HSD/1/162}. You explain here that you obtained an average historical beta across days of the week and you say in the last sentence of 435: "There is no strong theoretical basis to prefer any one seven-day period versus another." You are stating that because you are aware that what you've done by averaging the five days of the week is highly unusual, aren't you? A. So I would -- highly unusual. So what I would say is that it's not employed by numerous people, but it is the 100 Transcript Day 8, page 52 line 3-page 54 line 15. 101 Transcript Day 8 page 55 line 22-page 57 line 1. correct way to do it, and, in fact, in revisions of my own work, I'm now stating that one should look at this, for the following reason, which is -- Q. I'll come on to ask you about why. A. I would really like to finish this answer, sir. I was in the middle of it. And the reason that's the case is just what we were talking about earlier, which is that betas are estimated with error. And, in fact, I talk in my chapter, which we talked about earlier, that when you estimate betas, you would like to have as many different ways to cross-check it as possible. Perhaps the ideal way to cross-check it is with the exact same company. So the company returns for Nord Friday to Friday can be cross-checked by the Nord returns Wednesday to Wednesday, Tuesday to Tuesday. So the cross-checking it there makes a lot of sense. By averaging across all of these observations, we are actually reducing the standard errors around the beta because we have multiple draws. And therefore, averaging it gives us a much better estimate and reduces the error around our estimate of beta. We get a better estimate of beta.... Q. So beta has been commonly used for, what, 50 years or so, Professor? A. Yes. Q. And in those 50 years, where we have seen extensive research into beta, no one -- and I'm putting to you this -- no one has ever written an article or any other publication that suggests that one should average five sets of weekly returns. That is right, isn't it? A. I'm not sure that that would be a topic for an academic journal, but certainly it is the case that if one thinks a priori from both a theoretical and an empirical point of view -- Q. Sorry, let me have an answer to the question before we have another lecture. That is right, isn't it? A. I have not seen such an article but, as a tenured finance professor at Harvard Business School, who has written extensively on the topic, one wants to triangulate the cost of capital to make sure that you have the right number in this case, there is no reason not to use a beta on another day. Therefore, it makes sense that averaging would give us a better estimate of

¶ [197] In the course of the hearing I saw more immediate logic to Professor Gompers’ approach to taking steps, as it were, to neutralize randomness than in Professor Fischel’s contention that random differences in weekly data depending on the last week day chosen should be ignored. On reflection, however, it is difficult to avoid drawing the inference that Professor Gompers’ innovative and elaborate approach to calculating beta in this case would not have been deployed if the Friday to Friday data generated by Bloomberg had not happened to produce a higher figure than the average of data for weeks ending on other days of the week. He used data published by the Center for Research in Security Prices (“CRSP”), which itself provided data for the last day of the week (like Bloomberg) and indicated a beta figure of 1.24, far closer to the Bloomberg figure than the weekly average the Dissenters’ Expert calculated for himself. Professor Fischel also suggested that the use of the CRSP figures reflecting a combination of exchanges rather than the S&P 500 market proxy he used was unusual, but had no material impact on the result in the present case103.

¶ [198] I accept the approach of Professor Fischel which I find to be more orthodox and straightforward and I approve his adoption of a beta of 1.3 subject to deciding whether or not a ‘Blume adjustment’ is required. The need for a ‘Blume adjustment’

¶ [199] Professor Gompers in his Expert Report explained why such an adjustment was in his view required: “436...I next computed Nord Anglia’s CAPM historical adjusted beta by applying a two-thirds weight to the average CAPM historical beta of 1.06 and a one-third weight to the market beta of 1.00. This adjustment is consistent with Bloomberg’s beta calculation methodology, and it accounts for the expected future movement of betas towards a value of one (which has been empirically documented in the academic literature by the economist Marshall Blume), and is consistent with the notion of computing the cost of equity in a forward looking context. The adjustment yielded a CAPM adjusted beta of 1.04 based on the historical beta estimate of 1.06...”

¶ [200] Professor Fischel in his Supplemental Report firstly questions the support provided by an article Professor Gompers relies on to justify this approach104. He then proceeds in 102 Transcript Day 10 page 159 line 21 —page 161 line 4; page 164 line 8-165 line 4. 103 Supplemental Report, page 84 note 261. 104 Robert C Klemkosky and John D. Martin, ‘The Adjustment of Beta Forecasts’, 30 The Journal of Finance (1975) 1123-1128. the same paragraph to justify not applying the Blume adjustment on the following grounds: "93...Blume’s explanation does not support adjusting the Company’s historical beta downwards (as Professor Gompers does) because the Company’s new projects (i.e., new China Bilingual schools) are likely to be riskier than its existing projects (i.e., established premium international schools)."

¶ [201] I found that Professor Fischel’s criticism of Professor Gompers for relying on the Klemkosky and Martin article was effectively neutralised by Mr Bompas QC in cross-examination105. Professor Fischel agreed that he had not proffered the reason he advanced in his Supplemental Report for using an unadjusted historical beta. This was in part, he explained, because in his view it is more usual to use an unadjusted beta than an adjusted beta106. As regards his opinion that the Company’s future risk profile was a circumstance which justified not making an adjustment to the historic beta figure, the following critical cross-examination took place: "Q. And when you are arriving at your beta, what you are looking for...I mean, you can perhaps explain the difference to me. What you are looking for is systematic risk, not idiosyncratic risk. Idiosyncratic risk is the particular risk to do with the company. A systematc risk is really a more general correlation between the company and the area in which it works. Q Is that fair? A. That's fair. Q. So that the idiosyncratic risk shouldn't be relevant for the point at hand. A. It may or may not be."107

¶ [202] I reject Professor Fischel’s argument that no ‘Blume adjustment’ is required because of the Company’s idiosyncratic future risk profile. It remains to consider whether he is nonetheless right that, in effect, it is for the person contending for an adjustment to the historic beta (Professor Gompers in this case) to justify the need for an adjustment, because the need for such an adjustment is the exception rather than the norm. Mr 105 Transcript Day 8 page 72 line 23–page 77 line 15. 106 Transcript Day 8 page 79 line 25-page 80 line 5. 107 Transcript Day 8 page 81 line 23-page 82 line 9 Boulton QC carefully explored the underlying rationale for the 'Blume adjustment' a few days later in his cross-examination of Professor Gompers108: "Q. Let's look at what Marshall Blume himself said. Let's go to {F/10/1}. This is an article that you cite. So you can see it's 'Betas and Their Regression Tendencies' by Marshall E Blume. This is from 1975. If we go to page 2, first paragraph: {F/10/2} 'A previous study showed that estimated beta co-efficients, at least in the context of a portfolio of a large number of securities, were relatively stationary over time. Nonetheless, there was a consistent tendency for a portfolio with either an extremely low or high estimated beta in one period to have a less extreme beta as estimated in the next period. In other words, estimated betas exhibited in that article a tendency to regress towards the grand mean of all betas, namely one.' Q. Ie the whole market, Professor? A. You read that quite well, yes. Q. And this consistent tendency that Marshall Blume refers to is in respect of either an extremely low or a high estimated beta. That's right, isn't it? A. That's what at least this is talking about, the beginning of the paragraph, yes. Q. And your average beta of 1.06 before a Blume adjustment would not fit the definition of extremely high or extremely low, would it? A. It's close to 1. Q. I take it that's a yes, you are agreeing with my question? A. Yes. Q. And this article by the man behind the Blume adjustment, as we might say, goes on to examine why this might be the case. And we can see this at page 11. {F/10/11}. Picking it up, I think, at the bottom of the page, in the summary: 'In other words, companies of extreme risk -- either high or low -- tend to have less extreme risk characteristics over time. There are two logical 108 Transcript Day 10 page 176 line7-page 178 line 24. explanations. First, the risk of existing projects may tend to become less extreme over time. This explanation may be plausible for high risk firms, but it would not seem applicable to low risk firms.' There is no suggestion that that first logical explanation would apply in respect of a beta of 1.06 for Nord, is there? A. I don't quite know what you mean. Remember, we are measuring this off into the future. And again, it might be the case, but the first explanation is about the nature -- he is offering us one potential theory that the projects are closer to -- are less risky than the rest of the firm. Q. Yes. Well, and his second explanation, which I think you are touching on there, is: '... new projects taken on by firms may tend to have less extreme risk characteristics than existing projects.' So those are his two explanations. Neither of those could be said to apply to Nord in such a way that you would feel it necessary to apply a Blume adjustment from 1.06 towards the market beta of 1. They just don't apply in this context, do they? A. Again, I would think perhaps in the very long run, if we think about the beta, they might, and, as I said -- he doesn't talk about it in this paper, but it's certainly the case that others have talked about the idea that, given measurement error, a beta above 1 has the potential to be above 1, just because of noise in the data, and similarly below 1. And that clearly could apply here."

¶ [203] Professor Gompers was unable to convincingly refute the proposition, based on the writings of Marshall Blume himself, that the rationale for the Blume adjustment is to neutralize the effects over time of "extremely low or high estimated beta" based on data that showed that such extreme historic figures usually normalised over long periods of time109. This provides a coherent explanation as to why Professor Fischel did not feel the need to justify not using a 'Blume adjustment' in his DCF analysis and regarded recourse to the technique as more the exception than the rule. I find this dispositive of the issue. 109 Transcript Day 10 page 176 line 24 – page 178 line 24.

¶ [204] Accordingly, I find that no Blume adjustment should be applied to the historic beta figure calculated by Professor Fischel (based on, inter alia, the two year period he justifies in paragraph 91 of his Supplemental Report) of 1.30. Country Risk and/or Size Premiums?

¶ [205] Professor Fischel’s cost of equity calculation was done with only an “Equity Risk Premium” (10.28%), with only an additional “Size Premium” (11.79%), with only an additional “Country Risk Premium” (11.38%) and with both additional premiums (12.89%). However, in his Expert Report, he fairly admitted that both the Size Premium and Country Risk Premium were controversial and essentially left it to the Court to decide110. The Experts disagreed on the need for a Country Risk Premium but ultimately agreed on a Country Risk Premium Rate (if applicable) of 1.1%. Professor Gompers in his Supplemental Report argued against a Size Premium, in part citing Professor Fischel’s acknowledgment that the application of a Size Premium is controversial and in part opining that the real question was whether account had to be taken of any specific size risk factors111. Under cross-examination by Mr Bompas QC, Professor Fischel made no real attempt to persuade that the Court that a Size Premium was required on the facts of the present case: “A. No, here is the difference. He -- I think we both agree that there is a debate that the finance literature, to quote Professor Gompers, is inconclusive, that the academic -- the modern academic literature has been more critical of the existence of a size premium than was earlier the case. Practitioners, however, continue to use a size premium, and that's really the state-of-the-art, as it exists today, and I think Professor Gompers and I agree completely on that's the state-of-the-art. Professor Gompers took that present state of knowledge and decided a size premium was not appropriate, and I took the same thing and said there is no consensus, so, therefore, I calculated a cost of equity with and without a size premium and reported the results in my report.”112

¶ [206] This answer did not really engage with the true extent to which Professor Gompers in his Supplemental Report took issue with the general principle of a Size Premium, the very existence of which Professor Fischel himself acknowledged in his own Expert 110 Paragraphs 73-75. 111 Paragraphs 324-328. 112 Transcript Day 8 page 28 lines 7-22. Report was in doubt. I find that the case for the inclusion of a Size Premium in the costs of equity calculation has not been made out. Because of the predominant reliance I have decided to place on the DCF methodology of Professor Fischel, and having regard to my general sense of the appropriate ultimate fair value result, I find that no Country Risk Premium should be applied either. Professor Fischel opined that academic doubt existed about both forms of premium and I am not persuaded that they are appropriate to be used within the framework of the model that he constructed, Professor Gompers' "agreement" on the need for a Country Risk Premium notwithstanding Cost of debt

¶ [207] Professor Fischel in his Supplemental Report opined as follows: "95. Both Professor Gompers and I noted that Moody’s Investors Service had assigned Nord Anglia Education a long-term corporate debt rating of B1 and Standard & Poor’s had assigned the Company a corporate credit rating of B before the Announcement Date, and, for that reason, we both estimated the Company’s cost of debt based on the yield of corporate debt issued by companies with approximately the same debt rating. However, Professor Gompers used the effective yield of the ICE BofAML U.S. High Yield Index, which was 5.80% as of the Valuation date, whereas I used the effective yield of the ICE BofAML 10+ Year Single B U.S. High Yield Index, which was 8.15% as of the Valuation date, a difference of 2.358%.... 96.... In other words, Professor Gompers used an index comprised of below investment grade securities with remaining maturities of one year or more, whereas I used an index comprised of bonds rated B1 to B3 with remaining maturities of ten years or more... 97. Because ‘it is important to attempt to match the time period during which the cash flows are expected with that of the risk-free rate,’ Professor Gompers used the ‘yield on long-term U.S. Treasury Bonds’ in his calculation of the costs of equity. For the same reason, Professor Gompers should have used a long-term corporate debt rate to calculate the cost of debt...Because Professor Gompers used an index that included intermediate term debt with maturities of one to ten years, not just long-term debt with maturities of 10 years or more, he substantially understated the Company’s pre-tax cost of debts.”

¶ [208] Professor Gompers countered in his Supplemental Report that his own 5.80% Cost of Debt figure was more reasonable than Professor Fischel’s 8.16% for the following key reasons: (a) the “Fischel Bond Index” included only 26 bonds, 18 of which (69%) were either energy or retail companies. The “Gompers Bond Index” covered 731 bonds across 18 industries and was less biased; (b) the average years to maturity for the Company’s existing debt was only five years, which was closer to the six year average in the Gompers Bond Index; and (c) the Gompers 5.80% figure was closer to the Company’s historical cost of debt (5.88% at year-end 2015 and 6.88% per the May 31, 2017 audited financial statements). Professor Gompers’ figure was higher than the corresponding rate used by Houlihan Lokey in its fairness opinion (4.90%), a Goldman Sachs valuation (5.00%) and closer to the 6.20% used in the Project Bach Model than Professor Fischel’s estimate113.

¶ [209] Professor Gompers dealt with this aspect of his cross-examination in, at first blush, a convincing manner. He explained that the Cost of Equity was a long-term claim (“forever”) and it was therefore appropriate to match it with the cost of long-term risk-free investment. The relevant criterion when assessing the Cost of Debt was assessing what the duration of the Company’s debt was likely to be. The Company’s borrowing had been and was likely to be for no more than 5 years, so his chosen Index (covering bonds with an average 5 year term) was more appropriate than Professor Fischel’s. However, on reflection, the dismissive assertion that different periods of time logically applied when measuring the Cost of Debt and the Cost of Equity for the purposes of a DCF analysis did not deliver a knock-out blow to Professor Fischel’s insistence that the temporal element of equity and debt should be the same114: “Q. What you do disagree with then is if the average long-term debt of Nord was, say, six years or seven years, you would say that's too short? A. Yes, if you are trying to match the length of time for cost of debt, for example, to the length of time that you are using for a risk-free rate, where Professor Gompers and I both used 20 years, I would not use six years, if you are trying to match 20 years or longer because of the nature of a discounted cashflow model.”

¶ [210] Professor Fischel relied on two sources of support for his conclusion in his Supplemental Report. The first was Shannon P. Pratt and Roger Grabowski, ‘Cost of Capital, Applications and Examples’ (5th Ed., John Wiley & Sons, Inc, 2014) at 1206- 113 Paragraphs 318-321. 114 Transcript Day 8 page 103 lines 8-17. 1207. The relevant passage (only partially reproduced in the Supplemental Report\(^{115}\)) reads as follows: “...But many companies have taken advantage of low short-term interest rates and have borrowed short term for long periods of time. *The cost of debt capital should reflect the expected average of interest rates over a long period of time.* *If the current yield curve is upward sloping...that indicates that future interest rates are likely to be higher than current short-term rates.* You should consider using the long-term equivalent interest for bonds that would be comparably rated to the subject company…” [Emphasis added]

¶ [211] The quoted text supports the use of long-term rates, but only if there is a basis for believing that borrowing costs will likely rise over the long-term. It clearly suggests that the aim of the calculation is to predict what the actual likely borrowing costs will be over a long period of time; not that long-term borrowing costs should be applied to a company which is only likely to borrow on a short-term basis indefinitely into the future. I find the idea that one calculates the cost of debt over the period of predicted cash flows by reference to what the relevant borrowing costs are likely to be to be entirely consistent with the dominant aims of the valuation exercise. Using long-term debt rates simply because the relevant period is a long one makes no obvious sense.

¶ [212] The second passage relied upon by Professor Fischel, from Aswath Damodaran, *Damodaran on Valuation* (2\(^{nd}\) Ed., John Wiley & Sons, Inc., 2006) at page 65, does not provide any clear support for the relevant Expert opinion either. It supports the approach of using the rates of similarly rated widely traded bonds, but also suggests that “*recent borrowing spreads [may be used] to come up with a cost of debt*”. It is common ground that similarly rated and widely traded bonds may be used; but only Professor Gompers contends that it is desirable to match the indexed debt term with the actual debt term the Company has in the past used and is likely to use well into the future.

¶ [213] In the Company’s Written Closing Submissions, the case on cost of debt was summarised as follows: “*165. The other component of WACC is the cost of debt. Here the difference between the experts boils down to whether it is appropriate to estimate Nord’s cost of debt for inclusion in WACC on the basis of the short term or long term cost of debt. Professor Fischel says that it is appropriate to use the long term cost of debt, because this matches the time horizon of the cash flows to be discounted and is also consistent with the experts’ approach to the risk free rate (based on a 20 year government bond): {E/58.1/93} at paragraph 97. Professor Gompers uses the short term cost of debt which he said (incorrectly) matched*” \(^{115}\) Page 90 footnote 285. the duration of the Company’s existing debt: {E/72.1/115} at paragraph 320. Professor Fischel therefore derived his rate from bonds of 10+ years {HSD/3/67}; Professor Gompers used the same index but included all bonds {HSD/1/171}. As was put to Professor Gompers in cross-examination, this is like measuring the height of everyone in the room to estimate the height of the women in the room: {Day12/54:25} - {Day12/55:3}.”

¶ [214] I am unable to accept this submission. The evidence of Professor Gompers and the submissions of the Dissenters on this issue (i.e. whether future borrowing costs should be estimated based on the term of the loans the Company is likely to be subject to rather than the long-term nature of the estimation period) are far more cogent and coherent overall. In the Dissenters’ Written Closing Submissions, in addition to pointing out that Professor Fischel’s choice of index is unsupported by the academic materials he relies upon, they make the following central point: “398. What is relevant is matching the maturity of the debt issued by the Company: ‘... the debt rate you want has to match up to the maturity and the type of debt that the company has’. The Gompers Bond Index is consistent with the maturity profile of the Company’s actual long-term debt, which was approximately five years as of the announcement of the Merger; conversely the Fischel Bond Index is not comparable. The Company never issued debt with a maturity close to the average maturities in the Fischel Bond Index. That is unsurprising, given that a company can roll over maturing debt in the future, with the result that the maturities of its debt profile remain the same.”

¶ [215] I do accept the principle (explicitly supported by Shannon P. Pratt and Roger Grabowski, ‘Cost of Capital, Applications and Examples’) that if there is evidence that borrowing rates are likely to rise, that increase should be taken into account. The object of the exercise is to estimate the likely costs of debt for the subject company over the same period of time projected for estimating cash flows (perpetuity). However, Professor Fischel did not point to any such general market evidence. Under cross-examination by Mr Bompas QC, he agreed that he had possibly himself in previous cases calculated the cost of debt by reference to a company’s historical borrowing costs116. Accordingly, while I prefer Professor Gompers’ broad approach, and prefer his chosen Index to that of Professor Fischel’s, I do not consider it appropriate to accept his chosen rate of 5.80% uncritically without comparing it with the historic borrowing cost position as he accepts it is appropriate to do.

¶ [216] Professor Gompers justified his figure in his Supplemental Report in the following way 116 Transcript Day 8, page 100 lines 16-19. "321. Further, as explained in the Gompers Report, in addition to the estimate based on a ratings-based bond index, I also considered the cost of debt calculated based on the historical cost of debt disclosed in Nord Anglia’s financial statements and measured as interest expenses divided by the sum of long-term and short-term interest bearing debt. The historical cost of debt is 6.38% as of the most recent publicly available financial statements on May 31, 2017, and 5.88% as of fiscal year-end 2016 on August 31, 2016. Both estimates are significantly lower than the 8.16% used in the Fischel Report, and more consistent with the 5.8% used in the Gompers Report based on the bond index. Moreover, the costs of debt that I used in the Gompers Report is conservative compared to the estimates used in the Houlihan Lokey Fairness Opinion (4.9%) and in the valuation analysis by Goldman Sachs (5.00%), and broadly consistent with the cost of debt used in the Project Bach Model (6.20 %), all of which are lower than Professor Fischel’s estimate. 322. As an additional check on the appropriateness of the 5.80% cost of debt used in the Gompers Report, I examined the yield on Nord Anglia’s actual outstanding debt. Specifically, I calculated the weighted-average yield on Nord Anglia’s actual outstanding debt borrowings prior to the announcement of the Take-Private Transaction to be 5.21%. Therefore, the cost of debt used in the Gompers Report is well supported across a number of different sources, which are inconsistent with Professor Fischel’s estimate."

¶ [217] Three points arise in relation to these opinions which I initially found problematic. First, the historical cost of debt figures derived from the Company’s audited accounts are described as “estimates”, which diminishes their significance. Calculations based on historic figures in audited financial statements of the Company ought, prima facie, to be accorded a higher evidential status than what are genuinely only estimates as to future debt costs. Secondly, Professor Gompers characterises his 5.80% figure as conservative relative to other economic actors (Houlihan Lokey and Goldman Sachs), but departs from his predominant stance of placing reliance on the Project Bach Model to justify his ultimate DCF result. In this instance, the Project Bach Model figure of 6.20% is closer to the May 31, 2017-based historic debt figure of 6.38%. And, thirdly, while Professor Gompers’ analysis cogently demonstrates that his index-derived cost of debt figure is closer to the Company’s historic cost of debt figures, the estimates of other market participants and, indeed, the costs of the Company’s existing outstanding debt, he does not really explicitly justify privileging a figure derived from generic data over Company-specific historic borrowing information.

¶ [218] Having considered substituting the historic costs of borrowing rate of 6.38% for Professor Gompers’ 5.80 % figure, I consider I should defer to the consensus between the Experts that the appropriate measure to use is a rate derived from an index of comparably rated stock. Reference to historic borrowing is clearly regarded by Damodaran as a default reference point where the stock is not rated at all. The learned author explains the functional reason why the cost of debt is calculated in the following way which is instructive to the uninitiated117: 117 'Damodaran on Valuation' (2nd Ed., John Wiley & Sons, Inc., 2006) at pages64 -65. "The cost of debt measures the current cost to the firm of borrowing funds to finance its assets. In general terms, it should be a function of the default risk that investors perceive in the firm. As the perceived default risk increases, lenders will charge higher default spreads (on top of the risk-free rate) to lend to the firm... The most widely used measure of a firm’s default risk is its bond rating, which is generally assigned by an independent ratings agency. The two best known are Standard & Poors and Moody’s. Thousands of companies are rated by these two agencies, whose views carry significant weight with financial markets.... Many firms have bonds outstanding which do not trade on a regular basis. Since these firms are usually rated, we can estimate their costs of debt by using their ratings and associated default spreads... When there is no rating available to estimate the costs of debt, there are two alternatives: 1. Recent borrowing history..."

¶ [219] I accordingly approve Professor Gompers’ 5.80% cost of debt figure, which is a pre-tax figure. Applying the 31% statutory tax rate proposed by Professor Gompers and eventually agreed to by Professor Fischel118, to 5.80%, the post-tax cost of debt figure should be reduced to 4.00%. WACC

¶ [220] I will leave the parties to calculate precisely what impact the cost of debt figure I have approved has on the WACC calculation which should in all other respects be performed using Professor Fischel’s figures. My own best efforts suggest a figure in the region of 8.7%. Summary of findings on main DCF valuation issues in dispute

¶ [221] To summarise, I resolve the disputed DCF valuation issues as follows: (a) I approve Professor Fischel’s use of ‘Adjusted August 2017 Projections’ (taking into account the Company’s China Bilingual 5 Year Forecast); 118 Supplemental Report, paragraph 104. (b) I approve Professor Fischel’s terminal growth rate of 1.81%; (c) I approve Professor Fischel’s beta figure of 1.30 (with no Size Risk Premium or Country Risk Premium and without a ‘Blume adjustment’); (d) I approve Professor Gompers’ (pre-tax) cost of debt figure of 5.8% which results in a post-tax figure of approximately 4%; and (e) I find as a result that the WACC must be reduced to a figure to be agreed between the parties, but which I very roughly estimate as being in the region of 8.7%.

¶ [222] It follows that the appropriate DCF valuation must be adjusted upwards applying the reduced WACC (in an amount to be calculated by the parties but which I estimate to be in the region of 8.7%) to Professor Fischel’s valuation figures based on what he described as the Adjusted August 2017 Projections. I leave the parties to calculate the precise uplift which results from reducing the WACC from Professor Fischel’s low of 8.98% which produced an equity value per Share of US$41.32. My expectation is that this would result in a pro rata Share value in the region of US$44.

¶ [223] Such a valuation falls just above the Houlihan Lokey DCF range and at the top of the Professor Fischel Reports range.

¶ [224] It now remains to determine what the fair value of the Dissenters’ Shares should properly be, having regard to all the evidence, most notably the DCF valuation which I have just approved together with the Market Price and the Transaction Price. FINDINGS: THE FAIR VALUE OF THE DISSENTERS’ SHARES

¶ [225] Occasionally the critical issues in a case are best captured in one or two key submissions. In this case the overarching question in controversy was whether the Transaction Price or Merger Consideration of US$32.50 had been undervalued by more than 50% because the true intrinsic value of the Shares was US$76. In this case the best ‘jury point’ made on behalf of the Company came from Lord Grabiner QC, when he concluded his oral opening submissions as follows119: "In conclusion, I just want to say as follows, that if Professor Gompers is correct, then a large number of people were wrong. The special committee, including its advisers, they were all wrong. The analysts who published contemporary views, they got it all wrong. A majority of the minority shareholders at the date of 119 Transcript Day 1 page 84 line 18-page 86 line 2. the merger announcement -- they were people who either sold their shares or assented them to the merger -- they got it all wrong. The New York Stock Exchange and people investing on that exchange, they got it all wrong. Management who sold their shares, they were all wrong, which is interesting because they were insiders. CPPIB, the Canadians, which syndicated a proportion of the shares they had agreed to acquire in Nord; they got it all wrong, although they had done a good deal of their own scrutiny before coming to a decision as to whether or not to participate in the deal. Professor Fischel, he has got it all wrong -- but they would say he is parti pris but that will be a matter for your Lordship to decide. And then sophisticated investors, who had been, for example, in Fund IV -- and this is quite a nice point, that there were people who were investors in Funds III and IV, but in particular Fund IV, but who never came back for some more in Fund VI, which they were perfectly entitled to do, or would have been, but they, for some reason or another, decided that they would not participate in the post-merger syndication. These are very smart, sophisticated people. They obviously took the view that the kind of value that Professor Gompers suggests or contends for is simply not there. So essentially the whole world has got it wrong except for Professor Gompers. It's perfectly possible that he is right but in my respectful submission it is improbable in the extreme..."

¶ [226] The Dissenters, on the other hand, quoted from Oscar Wilde at the beginning of their Written Closing Submissions ("A cynic is a man who knows the price of everything and the value of nothing"120) as a riposte to the Company's primary reliance on the Market Price as indicative of fair value. Mr Adkin QC in his opening oral submissions cogently advanced the Dissenters' own main jury point121: "In this case, the majority shareholder, Baring, reached the well-informed view that the market significantly undervalued the shares in Nord and, quite lawfully, forced out the minority by way of a process in which it made clear that it would only sell its shares in the company to itself and its affiliates. Having done that, it has now performed, through the company it controls, a complete volte face and now 120 'Lady Windermere's Fan'. 121 Transcript Day 1 page 99 line 19-page 100 line 9. pretends to your Lordship, through the company, that the market got the value of the shares right all along and that the merger transaction was a sale process open to all which properly tested the market. This should be seen, we submit, for what it is, a piece of commercial opportunism, designed by Baring to play fast and loose with its obligation to pay fair value for what it has taken."

¶ [227] How I have evaluated these two seemingly starkly opposed views of the merits of the present case can perhaps be prefaced by reference to another Oscar Wilde quotation: "the truth is seldom pure, and never simple"122. As I have already indicated in recording my preliminary findings on the main controversial factual issues above, I have found merit in the main points advanced by both sides and essentially conclude that both sides are partly right and partly wrong, albeit leaning far more heavily towards the Company’s proposed valuation outcome. In summary: (a) I accept the Company’s broad submission that it simply beggars belief that the Sell Side acting as sophisticated investors would under-sell a valuable asset by more than 50%. There is no credible support for Professor Gompers’ valuation of US $76.51 as opposed to a Market Price of US$30.45 and a Transaction Price of US32.50; (b) I accept a much-diluted version of the Dissenters’ broad submission that the circumstances of the Merger Agreement call for a DCF valuation to be taken into account. This is principally due to (1) the involvement of Baring on both sides of the Transaction, (2) the existence of a controlling shareholder keen to sell yet eager to retain some indirect albeit significantly reduced interest in the Company, and (3) the fact that key market participants involved in and/or observing the Transaction (most importantly Houlihan Lokey, who gave a Fairness Opinion to the Company’s Special Committee) considered a DCF valuation was required; (c) I unequivocally reject any suggestion that the Transaction was less than a bona fide one and find no credible evidence of any intent, on the part of Baring or the Company, to confer a windfall on the Buy Side and to effectively cheat the Sell Side (which included Premier) on a massive scale. There were no material conflicts of interest which were not adequately dealt with through the Special Committee process which was deployed; (d) the Transaction Price, after critical scrutiny, provides important and credible evidence of fair value because the sales process was an arms’ length one and, despite concerns more of perception rather than 122 'The Importance of Being Earnest'. substance, reflected a transaction between a willing seller and buyer in circumstances where no serious competing bids were either made or shut out; (e) I find that the Transaction Price is a more reliable indicator of the fair value of the Shares than the Market Price because it was arrived at taking into account MNPI which may have resulted in the market undervaluing the Shares. Nonetheless there is a risk that because of a commercial desire to effect a sale as soon as possible, Baring’s preference for a sale to a known buyer as opposed to a complete outsider, and Premier’s blocking position, the Merger Consideration did not reflect the full intrinsic value of the Shares; (f) in a sale involving connected parties with no competing bidders and the ‘Sell Side’ clearly adopting a ‘bird in the hand’ stance, it would be surprising if the Transaction Price was above the intrinsic value of the Shares and more plausible that the Transaction Price was somewhat lower; (g) neither side has persuaded me that either the Transaction Price or, alternatively, a far higher DCF-derived valuation provides the sole solution to fairly appraising the fair value of the Shares. However: (1) the Company has persuaded me that the fair value is clearly much closer to the Transaction Price than the US$76.51 contended for by the Dissenters’ Expert, Professor Gompers, and (2) in light of the Houlihan Lokey DCF analysis (which suggested a mid-range value of just over $33 per Share) and the more rigorously tested Fischel (modified by Gompers) DCF analysis herein (which is indicative of a value over $40 per Share), I find that the fair value most likely lies somewhere between the Transaction Price and the DCF value approved by this Court.

¶ [228] The independent DCF analysis of Professor Fischel (which I have accepted with one notable modification which results in a slightly lower discount rate and, as a result, a slightly higher per Share value than he contended for at the top of his range) points to a value of just over $40 per Share. The independent DCF analysis of Houlihan Lokey (the technical validity of which the Dissenters’ Expert did not challenge) indicates an upper value of just under $40. This lends credibility to the DCF value I have reached, subject to concerns about the precise level of reliability of the August 2017 Projections, even as adjusted to take into account the China Bilingual 5 Year Forecast.

¶ [229] These concerns are ultimately not about the fairness with which the Projections have been prepared by honest and skilled professionals. It is more that, it seeming obvious to me that “sales patter” by insiders provide no reliable evidence as to value, the Company’s business prospects while largely stable were also materially unpredictable If short-term budgetary projections were indeed “stretching” ones which were never actually achieved, long-term projections must be viewed as even less reliable, as Mr Halder testified. The most insightful part of his testimony in this respect, which I accept, was the following passage (already quoted more fully above): “modelling risk weighting would not have materially improved our day-to-day decision making because the business was sufficiently strong to withstand individual schools materially underperforming...and as a business we had already decided to accept the large risks associated with doing business in many of the markets we were in...”

¶ [230] In my judgment, this evidence serves as a window through which the ‘real world’ in which the Projections were prepared can best be viewed, and supports the firm conclusion that the DCF valuation alone should not form the main foundation for appraisal in the particular circumstances of the present case where there is also other credible but not entirely reliable evidence of fair value. How then, can the Court take market indicators into account following a trial at which one Expert contended for reliance on market data alone and the other placed sole reliance on a DCF analysis?

¶ [231] Re Qunar provides a valuable example of how to resolve the conundrum of an imperfect DCF analysis and imperfect market evidence as to the fair value of a company’s shares in a section 238 appraisal case. In that case the Court was presented with a choice between: (a) a dissenters’ expert who contended for a DCF valuation producing a value nearly five times the merger consideration; and (b) a company’s expert who proposed a blended approach of 50% DCF and 50% market trading price, resulting in a value of less than the transaction price.

¶ [232] Nonetheless, Parker J adopted the blended approach of Ms Glass for the company because he was clearly unable to confidently rely exclusively on either a market price or a DCF analysis. This conclusion was reached against the background of the following findings: • “73. The DCF methodology can be an accurate measure of fair value. However, it depends upon the reliability of the models/projections, the various assumptions, and the validity of the inputs. Even a slight difference in these inputs can produce large variances. It also relies on subjective judgments to a large extent and can be easily manipulated by applying certain assumptions in the context of litigation As it is also something of an abstract concept, a cross-check may also be needed to bring a DCF valuation back to the 'real world' and prevailing commercial context. • "141. I have concluded that the share price of the Company in the particular circumstances of this case and the NASDAQ market at the time can reasonably be relied upon as good evidence of value. It therefore also provides a good cross check against the DCF outcome of fair value"; • "173. I do not accept that Mr Osborne's valuation is sustainable against the market price analysis of Ms Glass which I have accepted was reasonably conducted. Nor is it sustainable against the contemporaneous views of the analysts' community"; • "176. The logic of this finding is that a 100% DCF valuation is not the only appropriate valuation method in this case. I accept the blended approach adopted by Ms Glass as the best way of arriving at the fair value of the Dissenters' shares. The DCF and market trading approach both have advantages and disadvantages. Giving equal weight to both is in my view the most appropriate way to determine fair value in this case." • "188. Projections of this length were not prepared as a matter of course by the Company and it needs to be borne in mind that they were prepared in connection with approval sought for a statutory merger with Ctrip, its 94% majority shareholder at the time. D&P relied on Management Projections produced for the purpose and the Proxy Statement." • "189. I accept that the reliability of such projections may be affected by the purpose for which they are prepared and are in theory are susceptible to optimistic or more conservative treatment. The question in this case is whether they have in fact been prepared on a basis which means they are not reasonable, because they were prepared in order to fulfil another purpose which sought to reduce the value of the Company (i.e. were biased), or because they have been carelessly prepared, or contain obvious errors and are therefore unreliable or wrong for that reason." • "201. Ms Glass concluded, after having conducted her own detailed analysis, that the projections were by and large reasonable, save for one adjustment she made relating to income tax as a result of an exchange with Mr Zhu at the Management Meeting."

¶ [233] Although the justification for the company’s expert adopting a blended approach in that case is not explicitly set out, it seems clear that Parker J felt a combination of the market price and a DCF valuation was more reliable overall than a standalone valuation basis in circumstances where: (a) the management projections were considered reasonable, but it had been alleged they were unreliably conservative; (b) the market in the shares was regarded as providing good evidence of value, but it was said that the shares were grossly undervalued; (c) since the DCF valuation methodology was "something of an abstract concept, a cross-check may also be needed to bring a DCF valuation back to the 'real world' and prevailing commercial context"; and (d) Parker J concluded: "411. The consequence of a blended approach between a market trading valuation approach of the shares prior to the Merger and the DCF method leads in my view to a just and equitable outcome which will determine fair value for the Dissenters’ shares" [emphasis added].

¶ [234] A blended approach between the Transaction Price and the DCF valuation appears to me to lead to a just and equitable outcome in the present case where I have partially accepted each Expert’s criticisms of, inter alia, the Transaction Price and/or a DCF valuation as suitable sole valuation methodologies. But neither Expert has proposed a blended approach. In Re Shanda Games Limited, FSD 14 of 2016 (NSJ), Judgment dated April 25, 2017 (unreported), Segal J approved the following statement of principles which were uncontroversial in the present case, and to which I have already referred above: "85... in his opinion in Re Appraisal of Dell Inc 2016 WL 3186538 (Dell) Vice Chancellor Laster quoted with approval dicta in this issue from a number of other cases as follows... : "In discharging its statutory mandate, the Court of Chancery has discretion to select one of the parties' valuation models as its general framework or to fashion its own". M.G. Bancorporation, 737 A.2d at 525-26. "The Court may evaluate the valuation opinions submitted by the parties, select the most representative analysis, and then make appropriate adjustments to the resulting valuation". The court also may "make its own independent valuation calculation by . . . adapting or blending the factual assumptions of the parties' experts." M.G. Bancorporation, 737 A.2d at 524. It is also "entirely proper for the Court of Chancery to adopt any one expert's model, methodology, and mathematical calculations, in toto, if that valuation is supported by credible evidence and withstands a critical judicial analysis on the record." Id. at 526. "When . . . none of the parties establishes a valuation that is persuasive, the Court must make a determination based on its own analysis." [emphasis added]

¶ [235] I am satisfied that section 238 of the Companies Law confers jurisdiction on this Court which permits "adapting or blending" the approaches proposed by the Expert valuers. Should I give more weighting to either the Transaction Price or the DCF valuation or should each valuation measure be given equal weighting? I find this a very difficult question. Looking at matters in the round, it is on balance clear that my findings thus far overall justify the further finding that I consider that more weight should be given to the Transaction Price than to the DCF valuation. In other words, I have in substance used the DCF valuation as a cross-check for the Transaction Price rather than the other way around. I have not formed the view that the Transaction Price is so seriously lacking in credibility that the starting point should be a DCF analysis, with some account being given to the negotiated Transaction Price. In many important respects (save for rejecting his view that I should ignore a DCF valuation altogether), I have preferred the approach of Professor Fischel to that of Professor Gompers, who championed exclusive recourse to a DCF valuation approach. However I have accepted Professor Gompers' opinions on the unreliability of the Market Price and (in part) the unreliability of the Transaction Price.

¶ [236] I accordingly apply a 60% weighting to the Transaction Price and a 40% weighting to the DCF valuation approved by this Court (to be computed by the parties based on the findings recorded above). That should (and is intended to) result in a valuation which is modestly more than the Transaction Price of $32.50 but still within the potential range of DCF values calculated by Houlihan Lokey for the purposes of their Fairness Opinion. Minority Discount

¶ [237] Neither the parties nor the Experts positively proposed the application of a minority discount. Professor Fischel opined as follows in his Expert Report: "92. Finally, a DCF analysis...does not incorporate any valuation discount that may be applicable to minority holdings in a controlled company like Nord Anglia Education. However as noted above, I understand that the CICA has ruled that when a dissenting shareholder possesses a minority shareholding, it is to be valued as such."

¶ [238] Professor Gompers in his Expert Report opined in the salient part as follows: "422. The empirical evidence...is consistent with no minority discount and suggests a minority discount of no greater than 2% for the Dissenters shares in Nord Anglia if one were to be applied...Based on my review of the literature, I have identified no empirical studies that focus on the Cayman Islands, and as such, I relied on the values above for the United States and Hong Kong. These articles suggest a minority discount of up to 2% and potentially smaller and are consistent with no minority discount."

¶ [239] Mr Bompas QC in his closing oral submissions argued that having regard to the state of the evidence and the failure of the Company’s counsel to address the minority discount issue in their closing submissions, the Court should direct no discount at all123.

¶ [240] After the conclusion of the trial on December 20, 2019, the Judicial Committee of the Privy Council on January 27, 2020 delivered judgment in Shanda Games Ltd.-v-Maso Capital Partners Ltd et al [2020] UKPC 2. This was an appeal by dissenters’ against, inter alia, the Cayman Islands Court of Appeal decision holding that a minority discount should generally be applied in section 238 cases. In that case the experts agreed at trial that (a) fair value should be determined on a DCF valuation basis, and (b) if a minority discount was required, it should be 23%. Lady Arden (delivering the advice of the Judicial Committee) summarised the crucial findings as follows: "42. In the opinion of the Board, it is a general principle of share valuation that (unless there is some indication to the contrary) the court should value the actual shareholding which the shareholder has to sell and not some hypothetical share. This is because in a merger, the offeror does not acquire control from any individual minority shareholder. Accordingly, in the absence 123 Transcript Day 14 page 177 lines 5-22. of some indication to the contrary, or special circumstances, the minority shareholder’s shares should be valued as a minority shareholding and not on a pro rata basis. 55. It follows that the judge should not have held that fair value always means no minority discount (see, for example, judgment of the judge, para 93, second sentence). That could not be a bright-line rule to be applied in every case. Similarly, it was not open to CICA to hold that a minority discount should invariably be applied as a matter of law. The legislature’s direction is to find the ‘fair value’ of the dissenter’s shareholding. Because of the narrow scope of this appeal, the Board is not in a position to rule out the possibility that there might be a case where a minority discount was inappropriate due to the particular valuation exercise under consideration.”

¶ [241] The general rule and governing legal principle is that “in the absence of some indication to the contrary, or special circumstances, the minority shareholder’s shares should be valued as a minority shareholding and not on a pro rata basis.” However, if section 238 does not require “that a minority discount should invariably be applied as a matter of law”, in my judgment it must be for the parties and their experts to determine, in the first instance, what minority discount, if any, they consider is appropriate in each individual case. The starting assumption is that a minority discount should be applied; but this does not justify the Court determining that a particular minority discount should be applied as a matter of law without any evidential foundation.

¶ [242] In Shanda Games, the Court of Appeal (on March 6, 2018) applied a minority discount in an amount agreed by the respective experts, holding that a discount must be applied as a matter of law. The Privy Council upheld this decision, while clarifying that a minority discount might not be required in the particular circumstances of individual cases. The governing legal principle applicable to fair value appraisals under section 238 was articulated by Lady Arden as follows: “47...where it is necessary to determine the amount that should be paid when a shareholding is compulsorily acquired pursuant to some statutory provision, the shareholder is only entitled to be paid for the share with which he is parting, namely a minority shareholding, and not for a proportionate part of the controlling stake which the acquirer thereby builds up, still less a pro rata part of the value of the company’s net assets or business undertaking...”

¶ [243] When the Experts in this case prepared their Reports, the state of the law was that a minority discount was required as a matter of law. The Company adduced no expert evidence supporting the need for a minority discount on economic grounds, let alone proposed a specific discount figure. The Dissenters have adduced evidence through Professor Gompers to the effect that: (a) no minority discount is required on the facts of the present case, in effect because there is no economic justification for viewing a minority shareholder’s Shares in the Company as being financially impaired; (b) if a discount is legally required in general terms, the basis for the discount must be based on the value of control to the controlling shareholder or the negative value impact of the absence of control for the minority shareholder; and (c) if a minority discount is required as a matter of law to subtract the value of control, it would not (based on US and Hong Kong source material) exceed 2%.

¶ [244] Professor Gompers’ evidence that no minority discount is economically justified is not contradicted by any other evidence. On that basis alone, in my judgment I could perhaps find that no minority discount is required, the legal assumption in favour of a minority discount having been displaced. Should I without the assistance of expert evidence decide for myself that Professor Gompers’ uncontradicted economic views have no merit? I am required to form my own independent judgment and I find that the economic position is relevant in light of the governing legal position which trumps Professor Gompers’ view that economically speaking, “a company’s DCF value equals its fair value”124.

¶ [245] There is no judicial guidance as to how to determine the appropriate minority discount under section 238 where the figure is not agreed. In In re Shanda Games [2018 (1) CILR 352] (CICA), the experts were agreed as to the amount of the discount, if one was required; the legal requirement for one was what was in controversy. How one assesses the amount of the discount, assuming a legal requirement to value the actual shares held by the minority, was not considered. Martin JA’s central finding was as follows: “50 For these reasons, it appears to me that s.238 requires fair value to be attributed to what the dissentient shareholder possesses. If what he possesses is a minority shareholding, it is to be valued as such. If he holds shares to which particular rights or liabilities attach, the shares are to be valued as subject to those rights or liabilities. As a matter of mechanics, this can be done by adjusting the value that the shares would otherwise have as a proportion of the total value of the company; but failing to make such adjustments means that particular rights or liabilities will often be ignored, and the shares will be valued as something they are not.” 124 Expert Report paragraph 413.

¶ [246] This decision was upheld by the Privy Council with the addition of an important and practical caveat. This was to the effect that in some cases, where the amount of the minority discount was not agreed (as it was in Shanda Games), it might be open to this Court based on the specific valuation evidence before it to find that no discount at all was required. Again, the question of what factual issues might be relevant to deciding whether or not a discount was or was not required was not judicially considered.

¶ [247] In Re Qunar, the company’s expert admitted that she would not ordinarily apply a minority discount “assuming a publicly traded liquid security”, but proposed a 4.7% minority discount, presumably based on the legal assumption that one was required. The dissenters’ expert contended for a 0% discount. Parker J (at paragraph 406) found that “that the value of any discount to be attributed to the Dissenters being minority shareholders in the present case is nil”125 It appears to me that that Parker J concluded that there was no reliable evidence before the Court to support a finding that a minority discount in a specific amount or based on specific valuation considerations was required. He astutely adopted a legal approach the correctness of which has only been recently confirmed by the Privy Council in Shanda Games.

¶ [248] In the present case, I am bound to accept that a minority discount is required to take into account any value attached to control which cannot be attributed to minority shares. The Dissenters’ Expert has provided an opinion as to what that discount should be, assuming that the value of control should be assessed in the same way as in the US and/or Hong Kong. However I feel unable to fairly assess the validity of that opinion.

¶ [249] I find myself in a similar position to that of Parker J in Re Qunar of having no sufficient evidential basis for deciding the factual dimension of the valuation issues flowing from the legal requirement to value the Dissenters’ Shares as minority shares, rather than simply as a pro rata portion of the Company’s shareholding as a whole.

¶ [250] In future cases, in light of the guidance only recently received from the Privy Council in Shanda Games, expert witnesses and this Court will doubtless be able to develop on an incremental basis well-formed views on what economic factors are relevant to determining the need for and size of a minority discount under Cayman Islands law.

¶ [251] If, contrary to my primary findings, I were required to evaluate and reject Professor Gompers’ uncontradicted opinion that no minority discount is required, I would have applied a minority discount of 2% based on his also uncontradicted opinion that this is the maximum discount which could be required.

¶ [252] For these reasons I find that no minority discount should be applied as part of the valuation process in all the circumstances of the present case. 125 Paragraphs 403; 406. CONCLUSION

¶ [253] In summary, I find that the fair value of the Dissenters’ Shares should be computed by a blended approach comprised of (a) 60% of the Transaction Price of US$32.50, and (b) 40% of a DCF valuation based on the model constructed by Professor Fischel, with the WACC or discount rate adjusted downwards126 to a precise amount which I leave the parties (in consultation with their Experts, if required) to calculate. The reduction of the discount rate from Professor Fischel’s lowest proposed rate of 8.98% to what I very roughly estimate will likely be in the region of 8.7% should result in a total DCF pro rata share valuation no more than 10% above the upper limits of Professor Fischel’s range (US$41.45). I leave the precise result to be calculated by the parties.

¶ [254] The net result will be a valuation of the Dissenters’ Shares which falls within the DCF valuation range originally proposed by both Professor Fischel within these proceedings and by Houlihan Lokey, independent advisers to the Special Committee, in connection with the approval of the Merger Agreement. I have adopted a blended approach which was not proposed by either Expert because I found that the Transaction Price and, to lesser but significant extent, a DCF valuation provided more reliable indications of the fair value of the Shares than the Market Price.

¶ [255] This result reflects my findings that fair value may properly be found well above the Market Price of US$30.45 contended for by the Company, but far below the US$76.51 contended for by the Dissenters. Neither Expert positively proposed the application of a minority discount, despite being aware of a legal presumption that there should be such a discount under Cayman Islands law. Professor Gompers positively opined that no minority discount was required without being challenged by Professor Fischel. I accordingly find that no such discount should be applied.

¶ [256] I will hear counsel as to the terms of the final Order and costs, as well as in relation to any other matters arising from this Judgment, if required. THE HONOURABLE MR JUSTICE IAN RC KAWALEY JUDGE OF THE GRAND COURT 126 By reason of substituting Professor Gompers’ pre-tax Cost of Debt figure of 5.8% for Professor Fischel’s 8.16%.