McGraw-Hill Financial, Inc (now known as S&P Global, Inc) v Mitsub Pty Ltd (Trustee) - [2017] FCAFC 11

FEDERAL COURT OF AUSTRALIA
McGraw-Hill Financial, Inc (now known as S&P Global, Inc) v Mitsub Pty Ltd (Trustee) [2017] FCAFC 11

Appeal from: Application for leave to appeal and appeal: Mitsub Pty Limited v McGraw-Hill Financial Inc [2016] FCA 559

File number: NSD 842 of 2016

Judges: GILMOUR, FOSTER AND GLEESON JJ

Date of judgment: 2 February 2017

Catchwords: PRACTICE AND PROCEDURE – whether, by making an order in a class action and in relation to other class actions against rating agencies requiring the Chief Executive Officer of each of those ratings agencies to prepare an affidavit or affidavits specifying whether those agencies accept that certain findings made in other proceedings to which they were not parties have application to those class actions and, if not, why not, the docket judge erred in the sense explained by the High Court in House v The King (1936) 55 CLR 499 at 504–505 – whether, upon the true interpretation of the said order, the docket judge had power to make the said order

Legislation:
Australian Securities and Investments Commission Act 2001 (Cth), ss 12CB, 12DA, 12DB(1)(a) and 12DF

Civil Procedure Act 2005 (NSW)

Corporations Act 2001 (Cth), s 1041E and s 1041H

Federal Court of Australia Act 1976 (Cth), Pt VB, ss 31A, 37M(1), 37M(2), 37N(1), 37N(2) and 37P

Cases cited:
Articles cited:

Mitsub Pty Limited v McGraw-Hill Financial Inc [2016] FCA 559

ABN AMRO Bank NV v Bathurst Regional Council (2014) 224 FCR 1

Australian Coal and Shale Employees’ Federation v Commonwealth (1953) 94 CLR 621

Bathurst Regional Council v Local Government Financial Services Pty Ltd (No 5) [2012] FCA 1200

Decor Corporation Pty Ltd v Dart Industries Inc (1991) 33 FCR 397

Ex parte Bucknell (1936) 56 CLR 221

Expense Reduction Analysts Group Pty Ltd v Armstrong Strategic Management and Marketing Pty Limited (2013) 250 CLR 303

House v The King (1936) 55 CLR 499

Johnson Tiles Pty Ltd v Esso Australia Pty Ltd (2000) 104 FCR 564

Mallet v Mallet (1984) 156 CLR 605

Norbis v Norbis (1986) 161 CLR 513

Samsung Electronics Company Ltd v Apple Inc (2011) 217 FCR 238

Wingecarribee Shire Council v Lehman Brothers Australia Ltd (In Liq) (2012) 301 ALR 1

The CDO Market: Functioning and Implications in Terms of Financial Stability (Banque de France: Financial Stability Review: No 6 June 2005)

The role of ratings in structured finance: issues and implications (Committee on the Global Financial System of the Bank for International Settlements, publication No 23, January 2005)

Date of hearing: 6 July 2016

Registry: New South Wales

Division: General Division

National Practice Area: Commercial and Corporations

Sub-area: Corporations and Corporate Insolvency

Category: Catchwords

Number of paragraphs: 116

Counsel for the Applicants/Appellants: Mr SG Finch SC and Mr J Hewitt

Solicitor for the Applicants/Appellants: Clifford Chance

Counsel for the Respondent: Mr NC Hutley SC and Ms E Bathurst

Solicitor for the Respondent: Squire Patton Boggs

ORDERS

NSD 842 of 2016

BETWEEN:
MCGRAW-HILL FINANCIAL, INC (NOW KNOWN AS S&P GLOBAL, INC) (A COMPANY INCORPORATED IN NEW YORK)

First Applicant/Appellant

STANDARD & POOR'S INTERNATIONAL LLC (A COMPANY INCORPORATED IN DELAWARE)

Second Applicant/Appellant

AND:
MITSUB PTY LTD AS TRUSTEE FOR THE CHRIS CARROLL SUPERANNUATION FUND

Respondent

JUDGES:

GILMOUR, FOSTER AND GLEESON JJ

DATE OF ORDER:

2 FEBRUARY 2017

THE COURT ORDERS THAT:
1.The applicants be granted leave to appeal from the order made by Rares J on 18 May 2016 in proceeding NSD 1344 of 2015.
2.The draft Notice of Appeal received by the Court on 1 June 2016 stand as the Notice of Appeal in the appeal.
3.The appeal be allowed.
4.The order made by Rares J on 18 May 2016 in proceeding NSD 1344 of 2015 be set aside.
5.The costs of the appellants of and incidental to the said order as taxed or agreed be paid by the respondent.
6.The respondent pay the appellants’ costs of and incidental to the Application for Leave to Appeal filed on 1 June 2016 and the appellants’ costs of and incidental to the appeal.

Note: Entry of orders is dealt with in Rule 39.32 of the Federal Court Rules 2011.

REASONS FOR JUDGMENT
GILMOUR AND FOSTER JJ:

On 18 May 2016, the docket judge in proceeding NSD 1344 of 2015 (the proceeding below) made the following order (the 18 May order):
1.On or before 17 June 2016, the Chief Executive Officer of the respondents file and serve an affidavit identifying, after personally reviewing the files in this and each proceeding set out in the Schedule of Proceedings to this order, whether the respondents contend that the findings of:
a)the Court and the Full Court ([2012] FCA 1200 and [2014] FCAFC 65) in each of the Bathurst Council and related proceedings (first instance files NSD936/2009; NSD1073/2009; NSD1268/2010; and appeal files NSD501/2013; NSD502/2013; NSD503/2013; NSD504/2013; NSD505/2013; NSD507/2013; NSD508/2013; NSD522/2013; NSD523/2013; NSD524/2013) concerning the inappropriateness of the use of the Gaussian copula; and
b)the Court ([2012] FCA 1028) in the Wingecarribee proceedings (NSD2492/2007) as to the reports in 2005 by the Bank for International Settlements (“The role of ratings in structured finance: issues and implications” (CGFS publication No 23)) and Banque de France, (“The CDO Market: Functioning and Implications in Terms of Financial Stability”, Financial Stability Review, No 6 June 2005) concerning the inappropriateness of the ratings agencies’ methodologies;
in respect of rating structured products the subject of this and each proceeding set out in the Schedule of Proceedings address systemic events, do or do not apply to the respondents’ ratings methodologies used for each product in issue in each current proceeding and, if he or she believes that those findings do not so apply, explaining why not and why the court will need to deal with the issue of reliability of the respondents’ ratings that are impugned at a final hearing.
A schedule of proceedings was attached to the order. Seven sets of proceedings are listed in that schedule. In all but two of those proceedings, the defendant parties are McGraw-Hill Financial, Inc (formerly McGraw-Hill Companies, Inc which is now called “S&P Global, Inc”) (SPG), a New York company, and Standard & Poor’s International, LLC (SPI), a Delaware company. In proceeding NSD 417 of 2016, there is an additional defendant (Standard & Poor’s Financial Services LLC, a Delaware company). In proceeding NSD 1322 of 2012, the defendants are The Royal Bank of Scotland NV and McGraw-Hill International (UK) Limited, a company registered in England and Wales.
The docket judge in the proceeding below is the docket judge in some but not all of the proceedings listed in the schedule to the 18 May order.
On 20 May 2016, his Honour published his reasons for making the 18 May order (Mitsub Pty Limited v McGraw-Hill Financial Inc [2016] FCA 559).
SPG is the first defendant in the proceeding below. SPI is the second defendant in that proceeding. The primary judge referred to SPG and SPI together as “S&P” and we shall do the same in these Reasons.
S&P have sought leave to appeal from his Honour’s 18 May order. The order has been stayed by a member of the Full Court pending determination of S&P’s Application for Leave to Appeal and any subsequent Appeal. For the reasons which follow, we have decided to grant leave to S&P to appeal from that order and to allow the appeal. These are our reasons for doing so.
The 18 May order was not made as a sanction for non-compliance by S&P with any direction, order or Court-imposed obligation relating to the conduct of any of the proceedings listed in the schedule to the 18 May order. Nor was the 18 May order sought by any of the plaintiffs in any of those proceedings prior to it being proposed by the primary judge for the first time and without notice on 12 May 2016.
The evidence before us suggests that the Chief Executive Officer (CEO) of each of SPG and SPI lives in the United States. Neither CEO is a party to any of the proceedings referred to in the 18 May order. Although the position was not entirely clear when we heard S&P’s Application for Leave to Appeal, it appears that the CEO of SPG is not the same person as the CEO of SPI. That is, there are probably two individuals to whom the 18 May order is directed, not one. The precise facts concerning the identity of the CEO of each of SPG and SPI were not in evidence before the primary judge when he made the 18 May order.
THE PLAINTIFF’S CASE BELOW
The plaintiff in the proceeding below, Mitsub Pty Limited as trustee for the Chris Carroll Superannuation Fund (Mitsub), claims declaratory relief, damages or compensation, interest and costs against S&P in respect of alleged contraventions by those companies of s 1041E and s 1041H of the Corporations Act 2001 (Cth) (Corps Act) and alleged contraventions of ss 12CB, 12DA, 12DB(1)(a) and 12DF of the Australian Securities and Investments Commission Act 2001 (Cth) (ASIC Act) constituted by misrepresentations allegedly made by S&P to Mitsub in respect of a particular synthetic collateralised debt obligation (SCDO) called “the Blue Gum SCDO” and by unconscionable conduct in which those two corporations are said to have engaged. Mitsub also relies upon the common law tort of negligence.
In its Statement of Claim filed on 2 November 2015, Mitsub describes in some detail the structure and operation of the Blue Gum SCDO. This description is set out at pars 10–49 of the Statement of Claim. In general terms, as Mitsub pleads at par 10, a collateralised debt obligation (CDO) is a structured financial product whereby investors purchase the right to receive certain interest and principal payments from a number of underlying debt securities (including corporate bonds) and bear some risk in relation to defaults from those underlying debt securities.
At pars 50–57 of the Statement of Claim, Mitsub sets out the principal characteristics and performance factors of the Blue Gum SCDO. At pars 53 and 54, Mitsub alleges:
53.Credit event risk was the principal driver of the performance of the Blue Gum SCDO.
PARTICULARS
(a)Credit Events could reduce the funds available to earn BBSW interest on the SCDO Collateral to be paid to the Noteholders.
(b)Credit Events could reduce the funds available to the SPV to pay Coupons to Noteholders.
(c)Credit Events could reduce the funds available to the SPV to repay principal to Noteholders at maturity.
(d)Credit Events could result in a total loss of Noteholders’ capital. The number of Credit Events each tranche of the Blue Gum SCDO could withstand until total capital loss to Noteholders in that tranche was dependent upon several factors including inter alia:
(i)the level of subordination or credit support applied to the Blue Gum SCDO at the issue date by the arranger;
(ii) trading activity by the portfolio managers;
(iii)the weighting of each Reference Entity or Reference Entity as a percentage of the Reference Portfolio or Reference Portfolio; and
(iv)the recovery rate on the bond/s issued by the Reference Entity or Reference Entity which suffered the credit event (which is fixed at the issue date or variable as determined by the arranger and defined in the transaction documents applicable to the Blue Gum SCDO).
54.The number of Credit Events that could be expected to occur (as defined in the credit default swap) in the period to maturity of the Blue Gum SCDO which would require the SPV to make Protection Payments to the Swap Counterparty depended upon numerous factors, including:
54.1the creditworthiness of the issuers of credit instruments referenced by the credit default swap;
54.2the stability of credit markets generally and the availability of credit;
54.3the economic and financial environment impacting upon the operations of the issuer of credit instruments referenced by the credit default swap;
54.4the degrees of correlation between the issuers of credit instruments referenced by the credit default swap.
On or about 8 November 2005, S&P assigned an AA- rating to the Blue Gum SCDO (see par 58 of the Statement of Claim).
At par 59, Mitsub pleads:
59.The AA- rating is the second highest rating that may be assigned by S&P, with the indicator ‘-’ being an indicator of relative standing within the AA category. In assigning the Blue Gum SCDO a AA- rating, S&P intended to, and did, communicate and represent to investors or potential investors in the Blue Gum SCDO that:
59.1S&P had concluded that the capacity of the Blue Gum SCDO to pay Coupons to each of the Noteholders and the principal amount at the end of the term of the Blue Gum SCDO was very strong;
59.2S&P had concluded that the Blue Gum SCDO should be able to withstand a severe level of stress and still pay Coupons to each of the Noteholders and the principal amount at the end of the term of the Blue Gum SCDO;
59.3 those conclusions were based on reasonable grounds; and
59.4in reaching those conclusions and assigning a AA- rating to the Blue Gum SCDO being rated, S&P had exercised reasonable care and skill.
At par 60 of the Statement of Claim, Mitsub alleges that S&P represented to Mitsub and others that its credit ratings of SCDOs, including the Blue Gum SCDO, were objective, independent, uninfluenced by any conflicts of interest that might compromise S&P’s analytical judgment and reflected S&P’s true current opinion regarding the credit risks that SCDOs, including the Blue Gum SCDO, posed to investors. At pars 62–77 of the Statement of Claim, Mitsub describes the process undertaken by S&P in order to rate the Blue Gum SCDO. In those paragraphs, it makes the following allegations:
62.S&P’s rating of the Blue Gum SCDO represented its assessment of the probability of the Blue Gum SCDO defaulting and failing to pay any amount due to Noteholders. The rating addressed the likelihood of default but not the amount that may be recovered in a post-default scenario.
63.In order to assign a rating to the Blue Gum SCDO, S&P used a simulation model to attempt to predict the likely future performance of the Blue Gum SCDO, in terms of its ability to repay the principal amount at maturity and to make Coupon payments.
64.In order to assign a rating to the Blue Gum SCDO, S&P used a Monte Carlo simulation to estimate:
64.1the likelihood of Credit Events occurring in the Reference Portfolio and the proportion of scenarios in which individual tranches of the Blue Gum SCDO default;
64.2the distribution of losses across the tranches of the Blue Gum SCDO occasioned by Credit Events (the scenario default rate for each tranche); and
64.3the level of losses that a tranche of the Blue Gum SCDO must be able to withstand in the Monte Carlo simulations in order to achieve a given rating.
65.The assessment of the number of Defaults likely to be experienced among the Reference Entities in the period to maturity of the Blue Gum SCDO had two components:
65.1an assessment of the probability of a Reference Entity defaulting individually; and
65.2an assessment of the likelihood of events occurring in which more than one of the Reference Entities jointly default.
66.The first of those two components was determined by reference to historical probabilities of default identified in ratings tables for assets of the same class and same rating as the Reference Entities.
67.The second of those components was determined by a choice or an assumption of a probability distribution, known as a “copula” to be utilized in the model used to rate the Blue Gum SCDO and the parameterisation, or inputs to that distribution.
68.A copula is a mathematical concept used in statistics by which the interdependence of several variables subject to uncertainty (randomness) can be separated from the randomness of any one variable considered in isolation.
69.In the case of ratings for SCDOs such as the Blue Gum SCDO, the main variable of interest is the time to default (or survival) of the Reference Entities in the Reference Portfolio.
70.The copula distribution models the probability of multiple Reference Entities jointly defaulting and the time to default, rather than the likelihood of Reference Entities defaulting individually.
71.Where the chosen distribution copula is a Gaussian copula, the model input that defines that probability distribution is a correlation matrix.
72.The correlation matrix in the Gaussian copula model parameter models the degree of correlation among the Reference Entities, that is, the degree of probability that multiple Reference Entities will experience Defaults or Credit Events at or around the same time, having regard to the degree of their common exposure to systematic external events likely to cause Defaults.
73.For S&P’s modelling of the Blue Gum SCDO, a critical input to the model was an estimate, based on empirical data, of the correlation among the Reference Entities in the Reference Portfolio, meaning the degree of probability that multiple Reference Entities will experience Defaults or Credit Events at or around the same time, having regard to the degree of their common exposure to systemic external events likely to cause Defaults.
74.Correlation is likely to be higher among Reference Entities that operate in the same geographic region and the same industry. Correlation is likely to be lower among Reference Entities that are idiosyncratic, particularly Reference Entities located in different geographic regions and or different industries.
75.The model input for correlation will have a substantial impact upon the determination of when and how easily the attachment and detachment points for a SCDO tranche will be reached. That will in turn have a substantial impact upon the rating achievable for a SCDO such as the Blue Gum SCDO.
76.Increasing correlation increases the level of Defaults observed in the modelling process, which in turn has a negative effect on credit ratings for individual tranches of the SCDO. The higher the degree of correlation among the Reference Entities, the lower the rating.
77.Incorrect assumptions about default correlation can result in an inaccurate over-estimate of the creditworthiness of each tranche of the SCDO such as the Blue Gum SCDO.
At par 78 of the Statement of Claim, Mitsub alleges that S&P did not have reasonable grounds for concluding that the Blue Gum SCDO should be assigned a AA- rating. It provides detailed particulars of that allegation. Those particulars are in the following terms:
The methodology utilised by S&P to assess the Blue Gum SCDO could not reliably form the basis for the rating assigned to the Blue Gum SCDO pleaded in paragraph 78, having regard to the following.
(a) S&P’s use of the Gaussian copula model was inappropriate because:
(i)The behaviour imposed by the Gaussian copula for a group of Reference Entities is the equivalent to the asset returns for those Reference Entities being normally distributed (and correlated with each other). Normal distributions frequently underestimate the probability of extreme scenarios occurring.
(ii)The Gaussian copula model inadequately represents and materially underestimates the potential for joint severe events and the probability that assets will default in clusters because it underestimates the correlation out in the tail of distribution.
(iii)The use of the Gaussian copula model in S&P’s modelling of products such as the Blue Gum SCDO had the effect of underestimating the probability of each tranche of the Blue Gum SCDO experiencing a large number of Credit Events or Defaults and therefore underestimated the probability that each tranche of the Blue Gum SCDO would sustain losses.
(iv)S&P failed to use a “heavy-tailed” copula, such as the tcopula, which more accurately predicts the likelihood that extreme events will result in Reference Entities jointly defaulting. Heavy-tailed copulas assign a greater probability that extreme joint events will occur giving rise to multiple defaults than the Gaussian copula.
(v)When a heavy tailed copula is used to model a SCDO such as the Blue Gum SCDO with its inputs having been realistically and reliably determined by reference to empirical data, the heavy-tails indicate significantly increased risk of default in comparison to the Gaussian copula.
(vi)Use of a heavy tailed copula represents a realistic assessment of economic and market events that the SCDO such as the Blue Gum SCDO may experience, leading to joint defaults.
(vii)S&P’s use of the Gaussian copula model parameter meant that S&P’s model did not take into account the likelihood or probability of the Reference Entities experiencing extreme economic or market related events leading to multiple defaults among the Reference Entities, particularly having regard to the degree of correlation among them.
(b)S&P used estimates as to the number of individual defaults and also correlation in its modelling which were too low and underestimated the impact of individual defaults and correlation on the probability of default because S&P:
(i)Either lacked or failed to use empirical data that would have enabled it to accurately measure and estimate individual defaults and correlation among the Reference Entities.
(ii)Relied on the long term average of Defaults and default correlation data from a period of largely benign economic or market activity in which defaults were less likely to occur and in which the measure of correlation was likely to be lower.
(iii)Assumed that the long term average of defaults and default correlation derived during that period would apply across each period analysed in the Monte Carlo simulations.
(iv)Failed to isolate and use empirical data for Defaults and default correlation for periods in which default rates are materially higher than the long term average (during global recessions, financial crises and other periods of instability) which demonstrated higher degrees of correlation than the long term average utilised by S&P.
(c)S&P failed to test the performance of products such as the Blue Gum SCDO in exceptional but plausible stressed scenarios in which they were likely to experience multiple Credit Events or Defaults among the Reference Entities, leading potentially to a substantial loss of investor capital.
(d)S&P failed to test the performance of products such as the Blue Gum SCDO in scenarios in which it experienced a severe or extreme level of stress.
(e)S&P’s qualitative and quantitative analysis of SCDOs failed to have sufficient regard to the fact that:
(i)SCDOs were rated using modelling techniques that relied upon a large number of assumptions (particularly as to the number of individual Defaults, correlation and recovery rates) about the anticipated performance of the Reference Entities included in the Reference Portfolio that was extremely difficult to predict;
(ii)Minor adjustments to the assumptions and modelling inputs (including for defaults, default correlation and recovery rates) used by S&P in assessing the likely future performance of the SCDOs resulted in substantially different predictions about the likely creditworthiness of the SCDOs;
(iii)The high sensitivity of the SCDOs to economic and market risk factors including Credit Events meant that the likelihood of the SCDOs making their Coupons through to maturity and repaying capital at maturity was extremely difficult to predict; and
(iv)There was a substantial risk that predictions made by S&P about the likely future performance of the SCDOs could turn out to be incorrect.
(f)Each of the errors in S&P’s rating of the SCDOs described in paragraphs (a)-(e) above resulted in S&P materially underestimating:
(i)the probability of Defaults occurring among the Reference Entities during the period to maturity of the Blue Gum SCDO; and
(ii)the probability that each of the Noteholders would suffer losses in the value of their investment in the Blue Gum SCDO.
(g)Further particulars may be provided with expert evidence.

Thus, Mitsub alleges that the methodology used by S&P could not reliably form the basis for the AA- rating assigned to the Blue Gum SCDO by S&P. In addition, it alleges that the use of the Gaussian copula by S&P in its modelling was inappropriate for the reasons set out in subpars (i) to (vii) of the particulars provided in par (a) of par 78 of the Statement of Claim. That is to say, it is Mitsub’s case that the Gaussian copula should not have been used by S&P at all in order to rate the Blue Gum SCDO.
In probability theory and in statistics, a copula is a multivariate probability distribution for which the marginal probability distribution of each variable is uniform. Copulas are used to describe the dependence between random variables. The Gaussian copula is an example of a copula and is a complex mathematical formula.
In portfolio management and risk assessment, copulas are used to perform stress tests and robustness checks in order to assess the likely impact of extreme systemic events.
The use of the Gaussian copula in financial modelling in respect of credit derivatives has been criticised because, so the critics say, it exhibits a misleading or inaccurate representation of extreme events.
Mitsub also claims that, in its methodology, S&P used estimates of the number of individual defaults that were too low and thus underestimated the impact of individual defaults and correlation on the probability of default for the reasons set out in subpars (i) to (iv) of the particulars provided in par (b) of par 78 of the Statement of Claim. This is a criticism of the quality of the inputs included in the model.
S&P are criticised by Mitsub for failing to test the performance of products such as the Blue Gum SCDO in stressed scenarios.
S&P are also criticised for failing to conduct a qualitative and quantitative analysis that took into account appropriate considerations (as to which, see subpars (i) to (iv) of the particulars provided in par (e) of par 78 of the Statement of Claim).
Mitsub then alleges that the errors specified in subpars (a) to (e) of the particulars provided in par 78 of its Statement of Claim resulted in the material underestimations specified in subpar (f) of those particulars.
Mitsub alleges that S&P owed it a duty of care at all relevant times and that, as a result of S&P’s failure to exercise due care, Mitsub suffered loss.
At pars 90–94 of the Statement of Claim, Mitsub alleges that S&P did not have reasonable grounds for the rating which it ascribed to the Blue Gum SCDO and failed to take reasonable care in deriving and ascribing that rating. In those paragraphs, Mitsub pleads the following matters:
90.S&P ought reasonably to have been aware that the data, copula, information, assumptions and modelling inputs it relied upon in assigning the rating to the Blue Gum SCDO were inadequate and that it did not have reasonable grounds for the rating opinion it expressed with respect to the Blue Gum SCO, for the reasons pleaded and particularised in paragraph 78 above.
91.In the alternative, S&P was aware or ought reasonably to have been aware prior to publishing the rating for the Blue Gum SCDO that the deficiencies of using the Gaussian copula to model defaults were the subject of considerable discussion in the structured credit derivative industry and that substantial doubt had been cast upon the reliability of the Gaussian copula in modelling defaults for SCDOs which in turn cast doubt upon the reliability of rating which depended upon the use of the Gaussian copula. That was material information which S&P failed to disclose when it published its rating for the Blue Gum SCDO.
PARTICULARS
Further particulars will be provided with discovery and/or expert evidence.
92.Further, S&P ought reasonably to have been aware that minor adjustments to the assumptions and modelling inputs it relied upon in assigning its rating to the Blue Gum SCDO would have resulted in the AA- rating not being achieved.
93.S&P knew or ought to have known at that time that it assigned the Blue Gum SCDO rating, that the fact of that rating would be communicated to a class of persons in Australia who were potential buyers of the Blue Gum SCDO (which included the Applicant).
94.The S&P AA- Representations were false and misleading and/or misleading and deceptive and/or or misleading in a material particular because:
94.1S&P did not have reasonable grounds for reaching the conclusion that the Blue Gum SCDO should be assigned an AA- rating;
94.2the assignment of the AA- rating pleaded at paragraph 58 above to the Blue Gum SCDO was not the product of the exercise of reasonable care and skill;
94.3no reasonably competent ratings agency would have assigned a AA- rating to the Blue Gum SCDO, with the data and information that was obtained by S&P as a result of its modelling processes for the Blue Gum SCDO.
PARTICULARS
The particulars of paragraphs 78 are repeated.
In addition, at pars 97–99, Mitsub alleges that, contrary to the representations made by S&P to Mitsub, the rating which S&P ascribed to the Blue Gum SCDO was not independent.
At the heart of Mitsub’s case against S&P is its allegation that S&P’s use of the Gaussian copula model in S&P’s rating modelling was inappropriate. It contends that, by using the Gaussian copula model in making its rating of the Blue Gum SCDO, S&P failed to take appropriate account of the true probability of extreme scenarios occurring. Mitsub contends that the use of the Gaussian copula led to a serious underestimation of the potential for joint severe events and a serious underestimation of the probability that assets would default in clusters. Mitsub also challenges the suitability of certain inputs made by S&P into its model; the quality of the testing of S&P model undertaken by S&P; and, ultimately, the worth of its qualitative and quantitative analysis in respect of the Blue Gum SCDO.
We are told that, in all but one of the proceedings referred to in the 18 May order, substantially the same allegations are made against S&P in relation to the SCDOs the subject of those proceedings.
SOME MATTERS OF CONTEXT
On 20 December 2007, the Wingecarribee Shire Council and other Councils commenced a proceeding in this Court against Grange Securities Limited and others (the Wingecarribee proceeding).
Neither SPG nor SPI was a party to the Wingecarribee proceeding.
Judgment in the Wingecarribee proceeding was delivered by Rares J on 21 September 2012 (Wingecarribee Shire Council v Lehman Brothers Australia Ltd (In Liq) (2012) 301 ALR 1 (Wingecarribee)).
On 27 August 2009, the Bathurst Regional Council (BRC) commenced a proceeding in this Court against Local Government Financial Services Pty Ltd (and later others) (the BRC proceeding) for damages or compensation for losses suffered by the BRC flowing from defective ratings attributed to certain collateralised securities. That proceeding concerned a constant proportion debt obligation (CPDO) which, at that time, was a new form of product that had not previously been rated. The issues raised in the BRC proceeding were similar to (but not identical to) the issues raised in the Wingecarribee proceeding.
Justice Jagot delivered judgment in the BRC proceeding and two related proceedings on 5 November 2012 (Bathurst Regional Council v Local Government Financial Services Pty Ltd (No 5) [2012] FCA 1200 (the BRC judgment).
Neither SPG nor SPI was a party to any of the proceedings dealt with by the BRC judgment. However, McGraw-Hill International (UK) Limited was a respondent to all three of the proceedings dealt with by that judgment.
On 6 June 2014, the Full Court delivered judgment in the appeal from the judgment of Jagot J (ABN AMRO Bank NV v Bathurst Regional Council (2014) 224 FCR 1 (ABN AMRO)).
The proceeding below and two other extant proceedings concerning similar issues to those raised in the proceeding below were listed on 12 May 2016 before the primary judge.
Prior to the listing on 12 May 2016, the solicitors for S&P had written to the solicitors for the plaintiffs in two of the proceedings (including the proceeding below) which were, by then, listed on 12 May 2016, informing those solicitors that S&P proposed to seek summary dismissal of those proceedings pursuant to s 31A of the Federal Court of Australia Act 1976 (Cth) (FCA Act) upon the basis that all of the claims made therein were statute-barred. Interlocutory Applications seeking summary dismissal of those proceedings were filed on 10 May 2016 and supported by a Written Submission filed on 11 May 2016.
As at 12 May 2016, the proceedings listed for case management on that day had not all progressed to the same point of preparation. For example, in two of the proceedings no Defences had by then been filed.
At the case management hearings conducted on 12 May 2016, the primary judge raised the possibility of making an order along the lines of the 18 May order. This was the first time that S&P and its legal representatives were made aware of the primary judge’s intentions.
When the possibility of the Court making an order along the lines of the 18 May order was raised with the parties, Counsel then appearing for each of the relevant parties made oral submissions in support of their clients’ response to that possibility. Of necessity, however, the oral submissions made on 12 May 2016 were not fully developed, given that the parties had not received any notice of his Honour’s inclination to make an order along the lines of the 18 May order.
Later on 12 May 2016, after the case management hearings had concluded, the Associate to the primary judge emailed a form of order to the parties with a request that they settle that order and make any comments as to its form. The Associate went on to say:
His Honour’s intent is that, the Chief Executive Officer can file just one affidavit that deals compendiously with all of the proceedings to satisfy the orders applicable to all of them.
The next day (13 May 2016), which was a Friday, the legal representatives of S&P contacted the Associate to the primary judge and requested that his Honour allow S&P an opportunity to provide a brief Written Submission to the Court in relation to the proposed order. S&P indicated that it would provide such a submission by 11.00 am on 16 May 2016, which was the following Monday.
The lawyers for the plaintiffs in all of the proceedings objected to S&P’s having an opportunity to make such a submission.
The primary judge did not give S&P permission to file a Written Submission nor did he refuse to allow such a submission to be filed. The legal representatives of S&P filed a Written Submission on 16 May 2016 before the primary judge had responded to the request which they had made on 13 May 2016.
On 18 May 2016, the Associate to the primary judge sent another email to the legal representatives of the parties to which was attached the 18 May order. At the same time, the Associate informed the parties that his Honour proposed to publish his reasons for making the 18 May order on 20 May 2016.
THE REASONS OF THE PRIMARY JUDGE
At [1]–[3], of his Honour’s Reasons, his Honour said that:
(a)On 12 May 2016, at case management hearings of the proceeding below and of two other proceedings, he had proposed an order that had the purpose of requiring the defendants, by their CEO, to identify the real issues relating to the ratings methodology that S&P used in rating the various structured financial products the subject of eight representative proceedings currently before the Court (including the proceeding below);
(b)Following the same case management hearing, his Honour formulated the proposed order in Chambers and offered the parties an opportunity to comment on its form before his Honour settled the terms of the proposed order and made it formally;
(c)S&P subsequently (on 16 May 2016) filed a detailed Written Submission repeating and supplementing the objections to the making of any such order that it had already advanced in oral argument on 12 May 2016; and
(d)Ultimately, on 18 May 2016, his Honour made an order in Chambers substantially in the form of the proposed order which he had previously formulated.
At [4], his Honour referred to the BRC judgment and to ABN AMRO. In particular, his Honour noted that Jagot J had discussed the use of the Gaussian copula in relation to a CPDO in the following terms ([2012] FCA 1200 at [2556]):
I do not understand it to be suggested that S&P’s approach to default risk was of itself unreasonable. However, the expert evidence discloses that S&P would or ought to have known what was common knowledge in the industry, that modelling defaults using the Gaussian copula would seriously underestimate the potential for joint severe events because it “severely underestimates the correlation out in the tail of distribution”. The expert evidence also discloses that S&P must have known that its model modelled default risk separately from spreads which is not how markets actually work. Both facts would have benefited the performance of the CPDO as modelled by S&P. If there was room for qualitative analysis in the rating of the CPDO these kinds of facts should have caused a reasonably competent ratings agency to err on the side of caution when considering the results of the modelling. It will become apparent that S&P took the opposite course. (emphasis added)
The emphasis given to part of the above extract from the BRC judgment by the use of bold type was an emphasis added by the primary judge in the proceeding below and was not present in the judgment of Jagot J.
At [5], his Honour then referred to his own judgment in Wingecarribee at 26–27 [67]–[68]; 31 [83]–[84]; 225 [847] and 226 [851].
His Honour noted that the proceedings referred to in subpar 1(a) of the 18 May order are the BRC proceeding and ABN AMRO and that the proceeding referred to in subpar 1(b) of the 18 May order is the Wingecarribee proceeding.
At [5], his Honour went on to note that, in Wingecarribee, he had discussed aspects of ratings methodologies for SCDOs that had been examined by the January 2005 report of a working group of the Committee on the Global Financial System of the Bank for International Settlements: The role of ratings in structured finance: issues and implications (CGFS publication No 23) (the BIS 2005 report) and in an article published by the Banque de France in June 2005: The CDO Market: Functioning and Implications in Terms of Financial Stability (Banque de France: Financial Stability Review: No 6 June 2005) (the Banque de France article). Having referred to those documents, his Honour then quoted from Wingecarribee at 225 [847] and 226 [851]. In those paragraphs of Wingecarribee, his Honour said:
847The SCDOs offered a lower level of capital security than similarly rated ADI issued FRNs or corporate bonds, as the passage I have quoted from the Banque de France article in [83] explained. Critically, the SCDOs were susceptible to a qualitatively and substantively greater risk of unexpected loss because of their tranched structure and the products’ use of reference entities that were significantly lower rated. As the 2005 working group report explained:
Subordinated tranches, for example, can lose their entire value if losses in the asset pool are severe enough. This would not occur with an exposure to a straight bond portfolio, assuming that recovery rates are positive. Subordinated tranches will thus have higher unexpected loss than a bond portfolio with the same expected loss or probability of default. Given that the latter criteria (expected loss or probability of default) represent the one-dimensional risk indicators embodied in credit ratings, structured finance tranches can be riskier than investments in bond portfolios with equal ratings.
…
851A feature of the claim SCDOs as structured products was their particular susceptibility to extreme events that could lead to significant loss. As the 2005 working group observed:
The Working Group believes that risks associated with structured products may not have been fully grasped by some investors. Similarly, with consensus on “best practice” regarding the modelling of portfolio credit risk still lacking, “model risk” in instruments such as collateralised debt obligations (CDOs) is an issue for even the most sophisticated market participants. Use of structured finance instruments, together with the occurrence of worst case scenarios relating to mispriced or mismanaged exposures, might thus lead to situations in which extreme market events could have unanticipated systemic consequences. Given these issues and the fact that structured finance markets are still largely untested, continued growth in structured finance activity warrants ongoing central bank awareness.” (Emphasis in original.)
At [6], his Honour then correctly observed that S&P was not a party to the Wingecarribee proceeding and that there was no issue in the Wingecarribee proceeding as to the accuracy or reliability of S&P’s ratings. His Honour found that Grange Securities Limited (which later became Lehman Brothers Australia Limited) was the sole wrongdoer in the Wingecarribee proceeding.
At [7] of his Honour’s Reasons, his Honour summarised S&P’s submissions as follows:
In essence, S&P argued that there was no power in the Court to make such an order and that, if there were, it should not be made because:
Ÿthe chief executive officer would need to rely on hearsay, the affidavit would thus be of no probative weight and would not narrow the issues;
Ÿthe order could cause S&P to disclose legally professionally privileged information;
Ÿthere were less intrusive and less onerous means available to elicit a similar response, such as directions for S&P to file submissions;
Ÿthe earlier findings about the Gaussian copula and ratings methodology referred to in the proposed order did not relate directly to any of the structured financial products or SCDOs the subject of any of the eight current proceedings and did not decide that use of the Gaussian copula in rating those products was inappropriate or resulted in the assigned ratings being erroneous;
ŸS&P only had to show that there were rational grounds for the assignment of the relevant ratings;
Ÿthere was no evidence, or finding, in the Wingecarribee proceedings 301 ALR 1, that any of S&P’s ratings was incorrect;
Ÿthe applicants had not sought the proposed order;
Ÿthere was no limitation on how the contents of the affidavit might be used or what disclosure could be made of its contents;
Ÿthe proposed order was “more akin to a punitive sanction”, where there was nothing in S&P’s conduct that warranted censure. Moreover, in some of the eight proceedings, the pleadings had not yet closed;
ŸS&P had real and genuine grounds for defending its ratings methodology as evidenced in an affidavit made on 12 May 2015 by Cristina Polizu, a managing director and quantitative analyst of McGraw-Hill Financial Inc, the first respondent, in other proceedings brought by City of Swan that have recently settled; and
Ÿthe making of the proposed order would give rise to real and legitimate concerns as to whether I had prejudged the issues in the proceedings, including what will be a live issue as to the use of the Gaussian copula.
(Emphasis in original.)
At [8], his Honour referred to Pt VB of the FCA Act. In particular, his Honour summarised the effect of ss 37M(1), 37M(2), 37N(1), 37N(2), 37P(2), 37P(3)(a) and 37P(7) of that Act. His Honour also referred to Practice Note CM1.
At [9], his Honour then referred to the High Court decision in Expense Reduction Analysts Group Pty Ltd v Armstrong Strategic Management and Marketing Pty Limited (2013) 250 CLR 303 (Expense Reduction) at 323 [56] in order to reinforce the remarks which he had made at [8] to the effect that the Court is firmly in control of the conduct of civil proceedings. His Honour observed that the High Court had said at 323 [56]–[57] in Expense Reduction that it is the duty of the parties and their lawyers to assist the Court in furthering the overriding purpose of legislative provisions such as those embodied in Pt VB of the FCA Act and that that purpose may require a more robust and proactive approach on the part of the Court itself. Expense Reduction concerned various provisions of the Civil Procedure Act 2005 (NSW). Nonetheless, as the primary judge said, the reasoning employed by the High Court in that case is apt to be applied to the conduct of proceedings in this Court.

At [10]–[12] of his Reasons, his Honour said:
Having had, in some cases with Wigney J, responsibility for case management of these eight representative proceedings, I formed the preliminary understanding that a substantive issue common to all of them involves the assertion by the various applicants that S&P’s use of the Gaussian copula in its ratings modelling for each SCDO or other structured finance product in issue was flawed for the reason articulated by Jagot J in the passage I have quoted from her reasons (Bathurst [2012] FCA 1200 at [2556]). I also formed the preliminary understanding that her Honour’s reference to the Gaussian copula’s possible underestimation of “the potential for joint severe events” appeared to refer to a similar concept to that identified in the 2005 working group report as “situations in which extreme market events could have unanticipated systemic consequences” and “model risk” (Wingecarribee 301 ALR at 226 [851]) and the similar concept discussed in the Banque de France article (see 301 ALR at 31 [83], [84]).
Of course, I have not come to any clear view or understanding of these matters, and there is no evidence about them, in any of the current proceedings before me. Rather, I am seeking to understand how best to give appropriate directions in the future to enable the Court to deal with the real issues, many of which have superficially, at least, some common features. As Sheppard J (with whom Burchett J agreed on this point at 434) said in E I Du Pont de Nemours & Co v Commissioner of Patents (1987) 16 FCR 423 at 424, judges have power “to make, and continue to make, such directions as seem to them best suited properly and adequately to manage and direct the cases in their lists”. He said that the Court had a duty to make appropriate directions, even if necessary, against the will of the parties, that were adequate for and suited to the needs of the case. That inherent or implied power of the Court is reinforced by Pt VB and authorities such as Expense Reduction 250 CLR 303. I considered that it would be informative and important to understand, having regard to Jagot J’s finding about what S&P knew or ought to have known concerning what she found was the common knowledge of the issues concerning the use of the Gaussian copula in ratings modelling, what, if any, bearing her Honour’s finding and the two Central Bank reports, might have on the preparation of the eight current proceedings for trial.
Each of the SCDOs or structured products in issue appears to have experienced either some defaults or a wipe out because of extreme market events occurring from 2007, the effects of which are commonly known as the Global Financial Crisis. If (and I stress that at this point the question is based on a preliminary understanding of findings on other, but apparently related, issues after the two lengthy earlier trials) the Gaussian copula seriously underestimated the potential for joint severe, or extreme market, events, then it will be important to understand, at a broad, high level how S&P will defend its alleged use of the Gaussian copula to arrive at the ratings challenged by the various applicants.
At [13], his Honour then noted that the proceeding below and the other proceedings referred to in the schedule to the 18 May order had the potential to absorb a very large amount of Court time. His Honour then said (at [13]):
… I was concerned to ensure that the person within S&P with ultimate management responsibility, namely, the chief executive officer, explain at a high level, and of course without revealing any information that is protected by S&P’s right to maintain legal professional privilege, his or her understanding from actually considering the eight proceedings, as to whether the earlier findings referred to in the proposed order applied and, if not, what issues the Court would have to consider in relation to the reliability of S&P’s challenged ratings.
We pause to observe that, in his submissions before us, Senior Counsel for Mitsub contended that it was, at [12] and [13] of his Reasons, in particular, that his Honour explained why he had decided to make the 18 May order.
At [14], his Honour noted that Senior Counsel for most of the plaintiffs in the proceedings specified in the schedule to the 18 May order had submitted that the plaintiffs and their lawyers were not aware why, in substance, the finding of Jagot J as to the use of the Gaussian copula in ratings would not apply to S&P’s ratings of the SCDOs and any other structured products in issue in those proceedings. His Honour went on to say that the plaintiffs would be assisted by the filing of an affidavit along the lines of that which his Honour then had in mind.
At [15]–[17], his Honour said:
The litigation of those issues, if fully contested in respect of each rated product, is likely to be very expensive and time consuming both for the parties and the Court. I wished to be assured that the chief executive officer, as the person ultimately responsible for the decision to litigate them, had actually turned his or her mind to those issues and explained, at a high level, why the Court, having already spent considerable time and resources on different, but somewhat apparently (on my current understanding) related issues, should revisit those generated in the eight current proceedings involving the use of the Gaussian copula.
And, having considered these eight complex cases, the chief executive officer may well have his or her own insights as to how the real issues should be litigated. If there is some matter in the affidavit, when it has been made, that would make it appropriate for an order limiting its, or a part of its, use or publication, of course, S&P will be able to apply for orders to achieve such a purpose.
Conclusion
It follows that I reject S&P’s submissions. The purpose for which I sought the affidavit is to assist in the case management of the eight proceedings. It is not punitive or any kind of sanction. And, it is in no way based on any prejudgment about any of the complex and, as yet, protean matters potentially in issue. But, for the purpose of case management the Court can require a party, by its proper officer, to explain the reason why it needs to litigate particular issues in the proceedings and thus to require the Court and the other parties to devote resources to deal with those issues. That is why I made the order on 18 May 2016, that substantially reflected the form of the proposed order.
LEAVE TO APPEAL
In Samsung Electronics Company Ltd v Apple Inc (2011) 217 FCR 238 (Samsung) at 248–251 [25]–[37], the Full Court gathered together and reviewed the relevant principles governing the grant of leave to appeal in this Court.
As was the case in Samsung, the decision of the primary judge, if within power, was interlocutory, concerned a matter of practice and procedure and involved the exercise of a discretion. Of course, here the subject matter and type of order under challenge is very different from the order challenged in Samsung which was an interlocutory injunction.
At 248–249 [26]–[29], the Full Court said:
In this Court, it is well established that the relevant test (or “litmus test”) for whether leave to appeal from an interlocutory judgment will be granted, comprises the following two integers:
(1)Whether, in all the circumstances of the case, the decision is attended by sufficient doubt to warrant its being reconsidered by the Full Court; and
(2)Whether substantial injustice would result if leave were refused supposing the decision to be wrong.
(Decor Corporation Pty Ltd v Dart Industries Inc (1991) 33 FCR 397 at 398–400).
In Bienstein v Bienstein (2003) 195 ALR 225 at [29] (p 231), McHugh, Kirby and Callinan JJ said that:
The principles that govern the grant of leave to appeal are well established. An applicant for leave must establish that the decision in question is attended with sufficient doubt to warrant the grant of leave. The applicant must also show that substantial injustice will result from a refusal of leave to appeal.
The test for leave to appeal explained by the High Court in Bienstein v Bienstein is the same test as the Full Court had earlier articulated in Decor.
As the Full Court itself said in Decor, the test which it described is appropriate for the general run of cases. The test should not, however, be applied as if it were some hard and fast rule. Each case must be considered on its merits.
The Full Court then referred to the judgment of French J (as his Honour then was) in Johnson Tiles Pty Ltd v Esso Australia Pty Ltd (2000) 104 FCR 564 and to the High Court decision of Ex parte Bucknell (1936) 56 CLR 221 (Bucknell) at 225. In both of those cases, the Courts emphasised the need to consider the circumstances of each case. There are no hard and fast rules although there are guiding principles.
In Bucknell, the High Court said that the Court must examine each case and “… unless the circumstances are exceptional, it will not grant leave if it forms a clear opinion adverse to the success of the proposed appeal”. These remarks reflect the same sentiment as is found in the first limb of the test in Decor Corporation Pty Ltd v Dart Industries Inc (1991) 33 FCR 397 (Decor).
In Bucknell, the High Court also said that an interlocutory order affecting only the course of proceedings will seldom justify a grant of leave to appeal to that Court. The Court then gave some examples of such orders: Orders relating to interrogatories, discovery, particulars “… or like procedural matters …”.
In our judgment, the 18 May order is not in the same category of procedural orders as the High Court had in mind in Bucknell. It requires a number of parties (or, as S&P would have it, one or more individuals) to prepare evidence and to place that evidence before the Court as part of a compulsory process not consented to by the parties and not for the immediate purpose of a trial or other hearing but rather to inform the Court and those parties’ opponents of certain matters. The use to which that evidence may be put is not in the control of the parties ordered to prepare it and, as matters stand at the moment, is unknown. It seems to us that if, in all of the circumstances, the decision of the primary judge to make the 18 May order is attended by sufficient doubt to warrant it being reconsidered by the Full Court, then substantial injustice will be visited upon the party or parties or person or persons burdened by having to comply with that order.
As will be apparent from the reasons given in the next section of this judgment, we have formed the view that the first limb of the Decor test is satisfied in the present case. For the reasons which we have set out at [67] above, we are also satisfied that the second limb of Decor is satisfied.
S&P should, therefore, have leave to appeal from the 18 May order.
THE APPEAL
Introduction
S&P attacked the 18 May order on two broad bases: First, they argued that the learned primary judge did not have the power to make that order. Second, they submitted that, if his Honour did have power to make the 18 May order, his Honour erred in finding that the making of the order was appropriate in all the circumstances (ground 2). In effect, S&P’s second basis for attacking the 18 May order invoked the principles explained by the High Court in House v The King (1936) 55 CLR 499 (House v The King) at 504–505 per Dixon, Evatt and McTiernan JJ. In that case, at 505, their Honours said:
If the judge acts upon a wrong principle, if he allows extraneous or irrelevant matters to guide or affect him, if he mistakes the facts, if he does not take into account some material consideration, then his determination should be reviewed and the appellate court may exercise its own discretion in substitution for his if it has the materials for doing so. It may not appear how the primary judge has reached the result embodied in his order, but, if upon the facts it is unreasonable or plainly unjust, the appellate court may infer that in some way there has been a failure properly to exercise the discretion which the law reposes in the court of first instance. In such a case, although the nature of the error may not be discoverable, the exercise of the discretion is reviewed on the ground that a substantial wrong has in fact occurred. Unlike courts of criminal appeal, this court has not been given a special or particular power to review sentences imposed upon convicted persons. Its authority to do so belongs to it only in virtue of its general appellate power. But even with respect to the particular jurisdiction conferred on courts of criminal appeal, limitations upon the manner in which it will be exercised have been formulated.
See also Norbis v Norbis (1986) 161 CLR 513 at 518–519 per Mason and Deane JJ.
In Samsung at 252 [39], the Full Court said (after referring to House v The King):
In later cases, justices of the High Court have said that, in respect of appeals against decisions involving discretionary judgment, there is a strong presumption in favour of the correctness of the decision appealed from and that that decision should be affirmed unless the appeal court is satisfied that it is clearly wrong (see Australian Coal and Shale Employees’ Federation v Commonwealth (1953) 94 CLR 621 at 627 per Kitto J; and Mallet v Mallet (1984) 156 CLR 605 at 634 per Wilson J).
We accept that S&P must bring the present case within the principles explained by the High Court in House v The King in order to succeed in their appeal and must, therefore, demonstrate that his Honour’s decision to make the 18 May order was clearly wrong.
The Power to Make the 18 May Order
To a significant degree, resolution of the question of whether the learned primary judge had power to make the 18 May order will be determined by the Court’s interpretation of that order. In particular, a critical element in S&P’s contention that the Court lacked power to make the 18 May order is the proposition that the order should be interpreted as having been made against the CEO of each of SPG and SPI and of each of the respondents in the other proceedings listed in the schedule to the 18 May order. That is to say, it was S&P’s contention that the order was made against the individual or individuals occupying that position in each of the respondent corporations rather than against the corporations themselves. Upon the basis that this is the true interpretation of the 18 May order, it was then submitted that the learned primary judge had no power to make the 18 May order against a foreign individual who was not a party to this proceeding or to any of the other proceedings listed in the schedule to the 18 May order.
S&P submitted that:
(a)The general words of s 37P of the FCA Act do not provide power to make an order requiring a person who resides overseas and is not otherwise subject to the jurisdiction of this Court to do something outside Australia;
(b)If, contrary to the proposition stated in subpar (a) above, the general words of s 37P are sufficient to confer such a power, considerations of international law and comity were mandatory considerations that conditioned the exercise of the Court’s power and the primary judge failed to take into account any of those considerations;
(c)The general words of s 37P do not provide a power to make an order against a non-party;
(d)A further precondition to the exercise of the power conferred by s 37P(2) of the FCA Act is how the overarching purpose identified in s 37M(1) of that Act would be promoted by the making of the order. His Honour failed to give appropriate consideration to this precondition. In particular, the learned primary judge did not address why submissions made on behalf of S&P by their lawyers would not adequately address the case management objective which he was seeking to achieve; and
(e)The 18 May order is not truly an order which falls within the subject matter of s 37P(2) of the FCA Act viz giving directions about the practice and procedure to be followed in relation to the proceeding. By requiring the CEO of each of the relevant corporations to explain why the Court would need to deal with the issue of reliability of the ratings in question in the various proceedings covered by the 18 May order at a final hearing, that order required that person to address substantive matters affecting the adjudication of the respondent parties’ rights in the various proceedings. Such an order was not authorised by s 37P of the FCA Act.
The first answer which Mitsub made to these submissions was that, when properly construed, the 18 May order bound S&P and probably the respondents in the other proceedings listed in the schedule to the 18 May order, not the CEO of each of those corporations. In a detailed riposte, the applicants also took issue with the other arguments made by S&P in support of their proposition that the primary judge had no power to make the 18 May order.
We are inclined to think that, properly understood, the 18 May order was directed to SPG and SPI and perhaps to the other corporations identified as respondents in the proceedings listed in the schedule to the 18 May order. The question of the identity of the person or persons or entity or entities to whom the 18 May order is directed and precisely which persons or entities are therefore bound by that order is not without its difficulties and we appreciate that there is considerable force in the submissions advanced to the Court on behalf of S&P to the effect that the 18 May order was, in truth, directed to the CEO of each of SPG and SPI and perhaps of other corporations as individuals in circumstances where those individuals were not a party to any of the proceedings in question and were foreigners ordinarily resident overseas.
We are also inclined to think that if, in truth, the 18 May order was directed to foreign individuals who are not parties to the proceedings, it may well have been made without power. In that event, it would be liable to be set aside.
Because we have come to the firm conclusion that the discretion which the primary judge had decided to exercise in the present case miscarried, we do not propose finally to decide the questions raised by S&P’s contention that the 18 May order was made without power. For the purposes of this judgment, we will assume that that order was within power and will proceed to address ground 2 of S&P’s Draft Notice of Appeal.
Discretion
In their Draft Notice of Appeal, S&P provided detailed particulars of ground 2. Those particulars are in the following terms:
Particulars
(a)The primary judge failed to properly distinguish between the position of the chief executive officer of the Appellants, who is not a party to any of the proceedings, and the Appellants.
(b)The primary judge failed to take into account the caution required in making an order against a non-party.
(c)The primary judge failed to take into account the caution required in making an order against a person who is a resident of a foreign country and is not subject to the Court’s jurisdiction.
(d)The primary judge failed to have regard to how the making of the Order (directed personally to the chief executive officer of the Appellants) would promote the overarching purpose referred to in s 37M of the Federal Court of Australia Act 1976.
(e)The primary judge erred in failing to have regard to the risk of the appearance of pre-judgment in making the Order.
(f)The primary judge erred in making the Order on the mistaken premise that:
(i)findings were made in the Bathurst proceedings (at first instance and on appeal) concerning the inappropriateness of the use of the Gaussian copula;
(ii)findings were made in the Wingecarribee proceedings as to reports in 2005 by the Bank of International Settlements and Banque de France concerning the inappropriateness of the ratings agencies’ methodologies.
(g)The primary judge erred in acting without regard to the principle that a judge must not determine a case in a particular way because, in other proceedings, he or she or another judge has decided a matter a certain way.
(h)The primary judge erred in having regard to statements from counsel for the Respondent that the Respondent would be assisted by the affidavit contemplated by the Order given that:
(i) the need for the Order was not raised by the Respondent;
(ii)there was no evidence of any communications between the parties in which the Respondent sought information as to the basis for the Appellants’ defence on the issue addressed by the Order.
(i)The primary judge erred in failing to have regard to the prejudice to the Appellants that would or might be caused by the making of the Order.
(j)The primary judge failed to have regard to whether the case management objectives he was seeking to advance could be achieved by a different order such as an order that the Appellants (by their counsel) provide submissions.
(k)The primary judge erred in that the Order was not sufficiently clear or certain as to what the chief executive officer of the Appellants was required to do to comply with the Order.

Particulars (a), (b), (c) and (d) are all predicated upon the twin assumptions that the primary judge had power to make the 18 May order and that, upon the correct interpretation of that order, it was made against the CEO of SPG and also against the CEO of SPI. Given the views which we have expressed in relation to these assumptions at [73]–[78] above, we do not propose to consider the arguments based upon these particulars.
Particulars (e), (g), (h), (i) and (j) all involve contentions that the learned primary judge failed to have regard to a number of relevant material considerations and, in one case (particular (h)), had regard to an extraneous or irrelevant matter.
Particular (f) captures the proposition that his Honour misunderstood the effect of the findings in Wingecarribee and in the BRC judgment concerning the use of the Gaussian copula in the ratings models under consideration in those cases.
In both their written and oral arguments made to us, S&P developed their arguments by reference to the particulars given in their Draft Notice of Appeal.
S&P submitted that, even if the primary judge had power to make the 18 May order, his Honour did not take proper account of the fact that the order was made against one or more foreign individuals, none of whom is a party to the proceeding or to any of the other proceedings listed in the schedule to the 18 May order. For reasons already explained, we will put to one side arguments based upon an interpretation of the 18 May order to the effect that the order was made against one or more individuals.
Nonetheless, there is a very real argument as to the identity of the entity or entities or person or persons bound by the order. The order is ambiguous in this respect and is open to competing interpretations. It is left to the parties and their advisers to discern precisely who is bound by it. This is an unsatisfactory state of affairs.
It was next submitted by S&P that the primary judge had acted on a wrong principle when he made the 18 May order. S&P argued that a judge hearing a dispute must determine that dispute upon the basis of the material which is before him or her in that case and not because, in some other case, he or she or another judge in another proceeding has decided a similar matter in a certain way. By requiring the CEO to explain why certain findings or observations in Wingecarribee and the BRC judgment do or do not apply to the proceeding below and the other proceedings listed in the schedule to the 18 May order and why the Court will need to deal with the reliability of S&P’s ratings, the learned primary judge failed to have regard to the risk of the appearance of pre-judgment in making the order.
S&P also contended that the primary judge misunderstood the observations in Wingecarribee and the BRC judgment upon which he relied in making the 18 May order. S&P argued that the Gaussian copula was not the central issue in dispute between the parties in the BRC proceeding. Rather, so it was said, that case was principally concerned with the input assumptions that were used as part of the ratings process for a CPDO. Justice Jagot’s judgment was a lengthy one (3,723 paragraphs) and the only paragraph which addressed the Gaussian copula in terms was the paragraph cited by the primary judge in his Reasons for Judgment at [4]. Justice Jagot did not make any finding that the Gaussian copula should not have been used at all as part of the ratings process or that it necessarily produced a flawed rating. Rather, her Honour observed that there were limitations with the Gaussian copula that meant that a ratings agency was required to be cautious when considering the results of modelling using that copula. That is, the ratings agency had to consider the limitations of the Gaussian copula when conducting its qualitative and quantitative analyses based upon the output of any model used by it which has deployed the Gaussian copula. Such a finding cannot be extrapolated to qualitative and quantitative analyses applied to the outputs of another model altogether deployed to rate a different structured instrument. S&P also prepared a schedule of references to the evidence led in the BRC proceeding which they contended formed the basis of her Honour’s observations at [2556] of her judgment. That evidence suggested that some limitations on the use of the Gaussian copula were well known at the relevant time for the purposes of the BRC proceeding but that it was still being widely used at that time. The real issue with the copula, according to the evidence, was as her Honour had suggested at [2556] viz a reasonably competent ratings agency should have erred on the side of caution when considering the results of the modelling.
The reports relied upon by his Honour which had been referred to in his Honour’s earlier judgment in Wingecarribee similarly do not state that the ratings agencies’ methodologies were inappropriate in modelling portfolio credit risk. The key finding of the BIS Report was that “… risks associated with structured products may not have been fully grasped by some investors” (BIS Report at p 3). On the topic of CDO rating methodologies, the BIS Report noted that “… there is no current consensus regarding what ‘best practice’ is in terms of modelling credit risk in a portfolio context” (BIS Report at p 19 and also similar remarks at p 3). It further noted that models based upon Monte Carlo simulation “… seem to be emerging as the industry standard” (BIS Report at p 18). At p 55, the BIS Report referred to the Gaussian copula as “the most commonly used” copula. The BIS Report did not state that the use of the Gaussian copula in ratings agencies’ models was always or generally inappropriate nor did the primary judge go that far in Wingecarribee.
S&P also submitted that his Honour erroneously gave weight or too much weight to remarks made by Senior Counsel for Mitsub concerning the utility of making an order in the terms of the 18 May order. Senior Counsel had asserted that S&P had never put on any expert evidence or other evidence in any of the proceedings involving issues concerning its ratings methodologies as to what they say about the use of the Gaussian copula. S&P relied upon an affidavit of Cristina Polizu in order to challenge that assertion. We note that S&P’s reliance upon that affidavit led to a collateral dispute as to whether or not it truly amounted to expert evidence of the relevant kind. We do not think it necessary to resolve that dispute.
S&P also argued that the primary judge erred by failing to have regard to the prejudice occasioned to S&P and the other respondent parties listed in the schedule to the 18 May order which would or might be caused by the making of that order. The order requires the CEO or two CEOs to express belief on substantive legal matters. Those beliefs will inevitably be informed by legal advice. Thus, there is a real risk that, in complying with the 18 May order, the CEO of each of SPG and SPI and the other respondents will need to disclose privileged material. Although the learned primary judge referred to this problem, he did not solve it. Nothing in the order ameliorates the risk. In addition, S&P made the obvious point that there was nothing in the order itself which restricted the use that will or might be made of any affidavit filed in compliance with the order. While it is true that the control of the affidavit or affidavits and the use to which they may be put will be with the Court, the matter was left entirely at large.
Next, S&P submitted that his Honour failed to consider whether the case management objectives which he was seeking to advance could be achieved by a different order such as an order that S&P provide submissions directed to the question which was of interest to his Honour or be required to file some other issues document designed to tease out the matter of interest. S&P said that it was not necessary for the fair conduct of the proceedings or for the case management of the proceedings for the primary judge to have made the 18 May order without first providing S&P with an opportunity to consider whether the objectives which his Honour had in mind could be achieved by other means.
S&P also argued that, by expressly referring to the CEO in the 18 May order, the primary judge denied to S&P and the other respondents the opportunity to select the most appropriate person within the relevant corporations best able to provide the information sought by the 18 May order.
Finally, S&P submitted that the order was ambiguous in that it did not specify with sufficient particularity the task or tasks that the CEO of each relevant corporation was being required to do.
Mitsub responded to these submissions as follows:
(a)Although it had not made an application to the primary judge for an order in the terms of the 18 May order, Mitsub was entitled to take up the suggestions made by his Honour on 12 May 2016 and to press for an order along the lines of the 18 May order;
(b)The 18 May order is not ambiguous in any respect. In particular, it is not an order directed to individuals at all but rather an order directed to S&P (in the proceeding below) and to all of the respondents in the other proceedings listed in the schedule to the 18 May order;
(c)The 18 May order is quintessentially a matter of practice and procedure. It was made by a judge who is seised of a multiplicity of suits, all of which (with one exception) are dealing with the ratings of structured products. The ratings of such products are at the core of all of the proceedings with which his Honour and Wigney J are seised. The Full Court should not interfere with the 18 May order unless it is satisfied that the exercise of discretion miscarried and that there was prejudice or substantial injustice visited upon S&P by the making of that order. S&P have not discharged the burden of satisfying the Full Court of those matters. The primary judge was trying to avoid S&P taking up tactical positions in multiple actions. His Honour was trying to get a clear early statement of S&P’s position in respect of its ratings methodology. The explanation of the use of the Gaussian copula that will be brought forward in the affidavit or affidavits filed in compliance with the 18 May order will go a long way to enabling the Court and the other parties to understand exactly what S&P’s answer is to the allegation that the use of that copula was clearly inappropriate;
(d)No waiver of privilege will necessarily occur. The risk of a waiver of privilege occurring under compulsion can be solved by further case management in light of the affidavit or affidavits filed in compliance with the 18 May order;
(e)The argument now advanced by S&P that the 18 May order strips S&P of the power to select an appropriate deponent of the affidavit or affidavits which his Honour had in mind is a new argument and was not run before the primary judge. It should not be permitted to be run now. In any event, nominating the CEO of the two S&P corporations was a perfectly appropriate choice;
(f)The 18 May order was not premature. There was a mediation on the horizon and the information likely to be gleaned from a perusal of the affidavit or affidavits filed in compliance with the order was going to be useful at the mediation;
(g)The 18 May order is not ambiguous. Those who read it and seek to interpret it will be well informed about the issues in the proceeding below and the related proceedings and will be able to make sense of the order without difficulty; and
(h)There is no error of the kind explained in House v The King.
We think that the submissions made by S&P which we have summarised at [84]–[93] above are correct and we accept them. We would add the following observations.
While we recognise, as we must, that the 18 May order concerns a matter of practice and procedure, it nonetheless falls outside the types of practice and procedure order which the High Court in Bucknell held would not ordinarily be the subject of a grant of leave to appeal. The 18 May order is innovative, is not contemplated by the Rules of the Court and yet, subject to the Court having power to make such an order and subject to it being appropriate to do so, may be justified under the case management principles currently being developed by this Court in the modern context. But it is not a run of the mill procedural order.
Senior Counsel for S&P submitted that the order required the CEO of each of SPG and of SPI to review all of the commercial files relevant to the proceeding below and the other proceedings listed in the schedule to the 18 May order. Senior Counsel for Mitsub went further and submitted that not only was the CEO of those corporations required to review the commercial files, but he was also required to review all of the Court files listed in subpars (a) and (b) of par 1 of the 18 May order.
It is quite impossible to know whether the task that has been set for the CEO of each of SPG and SPI and of the other respondents is to review certain Court files, certain commercial files or both. Whichever interpretation one places upon the question of which files are to be reviewed, there is no doubt that the exercise required of each of the CEOs is a significant and burdensome one. This is not only because there will be a large amount of material for that person to review but also because of the other tasks that are or may be involved in that person complying with his Honour’s order.
For the reasons already explained, it is not possible simply to extrapolate for the purposes of the extant proceedings remarks or observations made by Jagot J in the BRC judgment and by his Honour in Wingecarribee. There are great dangers in seeking to do so.
It is one thing to require a party to explain its defence to the case sought to be made against it by the plaintiff, including, for example, in the present case, by requiring S&P to explain how it used the Gaussian copula in the model which it used to rate the Blue Gum SCDO; to explain why it considers that that use was appropriate; and to specify with particularity what steps it took to address the possible shortcomings of that copula when estimating the impact of extreme systemic events. It is quite another thing to require a party to study evidence and findings (or judge’s remarks) in entirely separate proceedings in order to place that party in a position to express a view as to whether the judge’s findings or remarks in those separate proceedings govern the issues in the instant case before the Court and, if not, as to the reasons why not. The flaws in such an approach are magnified if the parties in the instant case were not parties to the earlier proceedings and if the issues in the earlier proceedings are not the same as the issues raised in the instant case.
The 18 May order requires one or more CEOs of the relevant corporations to undertake the kind of comparison to which we have referred in the latter half of [100] above. In the present case, in the circumstances prevailing at the time when the 18 May order was made, this was a wrong exercise of the Court’s discretion.
In addition to ambiguities concerning the identity of the person or persons bound by the 18 May order and the scope of the expression “the files” when used in that order, there are other difficulties with the text of the order which render it utterly ambiguous.
First, as has already been submitted and accepted, the cases referred to in subpar (a) of par 1 of the 18 May order do not contain findings that the use of the Gaussian copula in the methodology adopted by ratings agencies to rate structured financial products is inappropriate. Thus, the CEO is being asked to search through the judgments referred to in that subparagraph in an endeavour to locate findings which were never made in those judgments. The order does not direct the CEO to any particular “findings” or remarks in those judgments and thus leaves to each CEO the decision as to what particular matters will comprise the basis for the comparison apparently required to be undertaken by the order. This inevitably results in the very real possibility that the basis for the comparison might differ from CEO to CEO and, in any event, might differ entirely from that which the primary judge had in mind when he made the 18 May order.
Second, a similar criticism can be made of the task required by subpar (b) of par 1 of the 18 May order.
Third, there appears to be a semantic difficulty caused by the language adopted in the first two lines after subpar (b) of par 1 of the order. At the hearing before us, Senior Counsel for Mitsub sought to solve that problem by inserting the word “to” immediately before the word “address” in the second line of text following the end of subpar (b) but this suggestion caused more problems than it solved. On one view of the requirements imposed by the order, the CEO is required to search out findings in the judgments referred to in subpar (a) and subpar (b) of par 1 which address the capacity of the Gaussian copula to estimate reasonably accurately the impact of extreme systemic events and then, if such findings can be found, to form a considered view on advice as to whether they apply to the proceeding below and to the other proceedings listed in the schedule to the 18 May order and, if not, why they do not so apply. In order to perform that task or those tasks, the CEO would need to ascertain the precise subject matter of any findings made in the earlier judgments which generally address the capacity of the Gaussian copula to accommodate extreme systemic events and then to make some kind of judgment about whether or not those findings can be extrapolated to the new proceedings. It is difficult to tease out this meaning from the text of the 18 May order and, of course, it may not be the meaning intended by the primary judge. In this most fundamental respect, the order is ambiguous.
We think that the 18 May order suffers from a number of serious ambiguities. Orders of the Court, so far as is possible, should be clear and unambiguous. These ambiguities provide a further independent basis for setting aside the orders.
CONCLUSIONS
For all of the above reasons, we are of the opinion that his Honour’s decision to make the 18 May order was wrong and clearly wrong within the meaning of that expression as used in Australian Coal and Shale Employees’ Federation v Commonwealth (1953) 94 CLR 621 at 627 per Kitto J and in Mallet v Mallet (1984) 156 CLR 605 at 634 per Wilson J.
The 18 May order must therefore be set aside.
There was no application before the primary judge by Mitsub for an order along the lines of the 18 May order. The genesis of that order is to be found entirely in the primary judge’s desire to obtain information from S&P and the other respondents which his Honour thought would assist the just, cheap and quick disposition of all of the relevant proceedings. In our view, the 18 May order should not have been made and there was no occasion as at mid May 2016 for any order along the lines of the 18 May order to be made. For these reasons, we do not propose to embark upon any purported re-exercise of the discretion which his Honour chose to exercise by making the 18 May order.
As S&P have been entirely successful in their Application for Leave to Appeal and subsequent Appeal, costs should follow the event.
There will be orders accordingly.

I certify that the preceding one hundred and eleven (111) numbered paragraphs are a true copy of the Reasons for Judgment herein of the Honourable Justices Gilmour and Foster.
Associate:
Dated: 2 February 2017
REASONS FOR JUDGMENT
GLEESON J:

I respectfully agree with the orders proposed by Gilmour and Foster JJ, and I generally agree with their reasons for allowing the appeal.

Concerning the grant of leave to appeal, in my view, substantial injustice would result if leave were refused (supposing the primary judge’s decision to be wrong) primarily because of the burden of complying with the order.
That burden has two aspects: the burden of swearing an affidavit addressing matters that are not identified with sufficient clarity to permit the deponent to know precisely what he or she was required to do to comply with the order; and the burden of substantial expenditure of resources to provide justification of the affected parties’ defences (or presumed defences) which was not demonstrated to be reasonably necessary to advance the aims of case management.
As to the form of the order, its difficulties are identified by the submissions made by S&P and summarised by Gilmour and Foster JJ at [87] and [88], and also by their Honours at [98] to [106]. In particular, I accept the submission made on behalf of S&P that the order is expressed upon mistaken premises that findings were made:
(1)in the BRC proceedings (at first instances and on appeal), concerning the inappropriateness of the use of the Gaussian copula; and
(2)in the Wingecarribee proceeding, as to statements in the BIS report and the Banque de France article concerning the inappropriateness of the ratings agencies’ methodologies.
As to the resources required to comply with the order, the order placed a significant burden on the CEO or CEOs. Those individuals can be assumed to have competing demands on their time, including multiple demands relating to the corporate governance of S&P. It would be a serious imposition to require the expenditure of that scarce time on compliance with an order that was made in error and that would probably require substantial effort having regard to the complexity of the issues that would need to be addressed by the affidavit or affidavits, and the solemn task of making an affidavit. This conclusion is reinforced by the fact that S&P are legally represented and there is no reason to doubt the capability of its lawyers to provide reliable information to the Court about the scope of S&P’s cases concerning the Gaussian copula and its methodologies.

I certify that the preceding five (5) numbered paragraphs are a true copy of the Reasons for Judgment herein of the Honourable Justice Gleeson.
Associate:
Dated: 2 February 2017

Appeal from:	Application for leave to appeal and appeal: Mitsub Pty Limited v McGraw-Hill Financial Inc [2016] FCA 559
File number:	NSD 842 of 2016
Judges:	GILMOUR, FOSTER AND GLEESON JJ
Date of judgment:	2 February 2017
Catchwords:	PRACTICE AND PROCEDURE – whether, by making an order in a class action and in relation to other class actions against rating agencies requiring the Chief Executive Officer of each of those ratings agencies to prepare an affidavit or affidavits specifying whether those agencies accept that certain findings made in other proceedings to which they were not parties have application to those class actions and, if not, why not, the docket judge erred in the sense explained by the High Court in House v The King (1936) 55 CLR 499 at 504–505 – whether, upon the true interpretation of the said order, the docket judge had power to make the said order
Legislation:	Australian Securities and Investments Commission Act 2001 (Cth), ss 12CB, 12DA, 12DB(1)(a) and 12DF Civil Procedure Act 2005 (NSW) Corporations Act 2001 (Cth), s 1041E and s 1041H Federal Court of Australia Act 1976 (Cth), Pt VB, ss 31A, 37M(1), 37M(2), 37N(1), 37N(2) and 37P
Cases cited: Articles cited:	Mitsub Pty Limited v McGraw-Hill Financial Inc [2016] FCA 559 ABN AMRO Bank NV v Bathurst Regional Council (2014) 224 FCR 1 Australian Coal and Shale Employees’ Federation v Commonwealth (1953) 94 CLR 621 Bathurst Regional Council v Local Government Financial Services Pty Ltd (No 5) [2012] FCA 1200 Decor Corporation Pty Ltd v Dart Industries Inc (1991) 33 FCR 397 Ex parte Bucknell (1936) 56 CLR 221 Expense Reduction Analysts Group Pty Ltd v Armstrong Strategic Management and Marketing Pty Limited (2013) 250 CLR 303 House v The King (1936) 55 CLR 499 Johnson Tiles Pty Ltd v Esso Australia Pty Ltd (2000) 104 FCR 564 Mallet v Mallet (1984) 156 CLR 605 Norbis v Norbis (1986) 161 CLR 513 Samsung Electronics Company Ltd v Apple Inc (2011) 217 FCR 238 Wingecarribee Shire Council v Lehman Brothers Australia Ltd (In Liq) (2012) 301 ALR 1 The CDO Market: Functioning and Implications in Terms of Financial Stability (Banque de France: Financial Stability Review: No 6 June 2005) The role of ratings in structured finance: issues and implications (Committee on the Global Financial System of the Bank for International Settlements, publication No 23, January 2005)
Date of hearing:	6 July 2016
Registry:	New South Wales
Division:	General Division
National Practice Area:	Commercial and Corporations
Sub-area:	Corporations and Corporate Insolvency
Category:	Catchwords
Number of paragraphs:	116
Counsel for the Applicants/Appellants:	Mr SG Finch SC and Mr J Hewitt
Solicitor for the Applicants/Appellants:	Clifford Chance
Counsel for the Respondent:	Mr NC Hutley SC and Ms E Bathurst
Solicitor for the Respondent:	Squire Patton Boggs

JUDGES:	GILMOUR, FOSTER AND GLEESON JJ
DATE OF ORDER:	2 FEBRUARY 2017