Model Risk Management in Wealth and Asset Management Josep-Lluis Perez, Banque de Patrimoines Privés Michael Hamp, RBC Global Asset Management Bill Bobey & Lars Meuller, CPP Investment Board June 04th 2015
Model Risk Management in Wealth and Asset Management Josep-Lluis Perez Banque de Patrimoines Privés June 04th 2015
Content 1. Introduction 2. Regulation 3. Model Risk Controls Business Lines 4. Model Risk Controls Risk Management 5. Governance Josep-Lluis Perez, MRM in Wealth and Asset management, Toronto, June 4 3
1. Introduction Models exist behind every decision wealth and asset managers take. The pricing models and trading strategies are key for the asset managers but also represent challenges; risks. Regulation is enhancing supervision on model risk management and is setting new requirements and standards that asset managers have to comply with. This presentation focus mainly on: - regulatory issues faced by asset managers - how risk managers can help achieving robust models - how the governance of the companies could be enhanced Josep-Lluis Perez, MRM in Wealth and Asset management, Toronto, June 4 4
2. Regulation Broad overview on European regulation and transposition into Luxembourg Law regarding the regulated investment universe for mutual funds (UCITS Directive) and alternative investment funds (AIFM Directive): UCITS DIRECTIVE Transposition into national Law AIFM DIRECTIVE European Directives Luxembourg Law From text Law to practical application CSSF REGULATION CSSF CIRCULARS references to Josep-Lluis Perez, MRM in Wealth and Asset management, Toronto, June 4 ESMA guidelines 5
2. Regulation In practice for model risk management: - Enhanced governance - CSSF Circular 12/546 - Sound risk management process - CSSF Circulars 11/512 and 12/546 - Independent risk management function - CSSF Circular 12/546 - Independent model validation - CSSF Circulars 11/512 (ESMA Guidelines 10-788) - Authority of the risk management function - CSSF Circular 12/546 - Experience, Education, Training - CSSF Circular 12/546 - Transparency, Disclosure CSSF Circulars 11/512 - Conflicts of interest policy - CSSF Circular 12/546 Josep-Lluis Perez, MRM in Wealth and Asset management, Toronto, June 4 6
2. Regulation Asset managers use models for pricing and trading strategies. In order to comply with current regulation asset managers must: - Disclose the pricing policy - Have written procedures - Determine on an ex-ante basis the type of assets that the company is able to price or the trading strategy that will be pursued Josep-Lluis Perez, MRM in Wealth and Asset management, Toronto, June 4 7
3. Model Risk Controls Business Lines An asset manager must : - Have the approval of the pricing committee on the models, parameters estimators and assumptions - Review the implementation of the models - Review the assumptions on a regular basis - Run independent checks on the estimated parameters Josep-Lluis Perez, MRM in Wealth and Asset management, Toronto, June 4 8
4. Model Risk Controls - Risk Management Main characteristics of the risk management function: - Independence - Qualified human and technical resources - Authority => Right to veto the validation of models Josep-Lluis Perez, MRM in Wealth and Asset management, Toronto, June 4 9
4. Model Risk Controls - Risk Management Independence: - Must be independent from the operational units - Its roles and responsibilities must be clearly defined and approved by the Board of Directors - Must report to Management and Board of Directors on a regular basis - Must have access to the Board of Directors Josep-Lluis Perez, MRM in Wealth and Asset management, Toronto, June 4 10
4. Model Risk Controls - Risk Management Qualified human and technical resources: - Must have the adequate education and experience to fulfill its obligations - As key persons of the company => Need to be assessed and approved by the Board of Directors - Must have the technical resources in line with their responsibilities=> the Board of Directors is responsible for allocating the budget to the internal control functions and the capital to the model risk Josep-Lluis Perez, MRM in Wealth and Asset management, Toronto, June 4 11
4. Model Risk Controls - Risk Management Authority: - Must have the right to veto any model - Must have access to all the information related to the models used, their implementation and assumptions used, and on any issue resulting in the use of such models Josep-Lluis Perez, MRM in Wealth and Asset management, Toronto, June 4 12
4. Model Risk Controls - Risk Management In practice: - Inventory of all the models used by the company - Classification of the risk models by severity and complexity - Review the back testing results - Review of the scenario analysis Josep-Lluis Perez, MRM in Wealth and Asset management, Toronto, June 4 13
5. Governance The governance framework: - Independent Directors - Corporate governance policy - Risk committee - Pricing/Trading strategy committee - Internal audit - Training programme Josep-Lluis Perez, MRM in Wealth and Asset management, Toronto, June 4 14
Model Materiality in Asset Management Michael Hamp, Ph.D. Vice President, Head of Risk Management June 2015
Model Risk in Asset Management Common Model Risk Framework Application to Asset Management Materiality of Investment Decision Models Case Study Implications Summary Best Practices 16
Model Risk Management Model Definition Model Inventory Model Risk Assessment Complexity and Materiality Model Validation 17
Model Risk Assessment Complexity Materiality differentiates model risk in asset management vs banks Materiality 18
Model Materiality at a Bank Valuation Models Regulatory Models Other Criteria high risk of loss, high reputational risk 19
Model Materiality at an Asset Manager Risk Principles for Asset Managers - 2008: Some models are relied upon for official calculations (i.e. valuations, fee calculations, etc.) and some are for internal, analytical purposes only. While vetting and review can be useful for all models, it is critical for the first category. Buy Side Risk Managers Forum & Capital Market Risk Advisors 20
Case Study AXA Rosenberg 21
AXA Rosenberg What Happened? Varying but similar reports Units error introduced in 2007 A risk model factor was underweighted by 100X or 10,000X Error discovered internally in June 2009 The error was corrected in September & October 2009 Rosenberg: Delayed correcting the error Failed to disclose the error to the firm s board or investors Instructed employees not to disclose the error 22
AXA Rosenberg What Happened? Factor weights were disclosed to clients Disclosure issues transformed the modeling error into fraud Actual impact of the error was never disclosed and may not be known Firm fined $25 million Firm agreed to compensate clients $217 million Barr Rosenberg barred for life from the investment industry 23
Disclosure Considerations Even in quantitative funds the PM must apply judgement PM judgement is a key protection for clients Disclosing specific model parameters removes PM judgement 24
Materiality for Asset Managers Official Calculations (valuations, etc.) Investment Decision Models: Evaluate the impact the model has on investment decisions Is the strategy highly quantitative? How complex is the model? How easily can the Portfolio Manager detect errors? 25
Summary Model Risk Framework for Asset Managers General Framework Model Risk Policy - Model Definition & Controls Model Inventory Model Risk Evaluation Materiality & Complexity Model Validation where Appropriate Asset Manager Considerations Model Materiality Definition Disclosure Considerations 26
This report has been provided by RBC Global Asset Management Inc. (RBC GAM) for informational purposes only and may not be reproduced, distributed or published without the written consent of RBC GAM. It is not intended to provide legal, accounting, tax, investment, financial or other advice and such information should not be relied upon for providing such advice. RBC GAM takes reasonable steps to provide up-to-date, accurate and reliable information, and believes the information to be so when printed. Due to the possibility of human and mechanical error as well as other factors, including but not limited to technical or other inaccuracies or typographical errors or omissions, RBC GAM is not responsible for any errors or omissions contained herein. RBC GAM reserves the right at any time and without notice to change, amend or cease publication of the information. Any investment and economic outlook information contained in this report has been compiled by RBC GAM from various sources. Information obtained from third parties is believed to be reliable, but no representation or warranty, express or implied, is made by RBC GAM, its affiliates or any other person as to its accuracy, completeness or correctness. RBC GAM and its affiliates assume no responsibility for any errors or omissions. All opinions and estimates contained in this report constitute our judgment as of the indicated date of the information, are subject to change without notice and are provided in good faith but without legal responsibility. To the full extent permitted by law, neither RBC GAM nor any of its affiliates nor any other person accepts any liability whatsoever for any direct or consequential loss arising from any use of the outlook information contained herein. Interest rates and market conditions are subject to change. A Note on Forward-looking Statements This report may contain forward-looking statements about future performance, strategies or prospects, and possible future action. The words may, could, should, would, suspect, outlook, believe, plan, anticipate, estimate, expect, intend, forecast, objective and similar expressions are intended to identify forward-looking statements. Forward-looking statements are not guarantees of future performance. Forward-looking statements involve inherent risks and uncertainties about general economic factors, so it is possible that predictions, forecasts, projections and other forward-looking statements will not be achieved. We caution you not to place undue reliance on these statements as a number of important factors could cause actual events or results to differ materially from those expressed or implied in any forward-looking statement made. These factors include, but are not limited to, general economic, political and market factors in Canada, the United States and internationally, interest and foreign exchange rates, global equity and capital markets, business competition, technological changes, changes in laws and regulations, judicial or regulatory judgments, legal proceedings and catastrophic events. The above list of important factors that may affect future results is not exhaustive. Before making any investment decisions, we encourage you to consider these and other factors carefully. All opinions contained in forward-looking statements are subject to change without notice and are provided in good faith but without legal responsibility. /TM Trademark(s) of Royal Bank of Canada. Used under license. RBC Global Asset Management Inc. 2015. 27
Portfolio Construction with Alternatives Bill Bobey & Lars Meuller CPP Investment Board June 4, 2015 28
Portfolio Construction Challenge Active Investment Strategies tend to exhibit at least one of Non-normally distributed returns Lagged equity exposure (i.e. illiquidity) Asymmetric systematic risk exposure Time varying systematic risk exposure Mixed data quality (i.e. frequency, short history, transparency to underlying positions) How do you address all of these when determining the composition of a portfolio of alternatives? One solution: use strategy pay-off profiles (w.r.t. a systematic risk such as equity) as building blocks to attain a target pay-off profile Bobey & Meuller, Portfolio Construction with Alternatives, Toronto, June 4, 2015 29
Strategy Examples: Quant EMN and FI Arb Data: Credit Suisse Hedge Fund Indices and MSCI World Equity Index 60-month rolling betas Pre-08 vs Post-08 upside (black) and downside (red) equity beta What are the conditional and unconditional correlations between these two strategies? What odel est specifies these relationships? Bobey & Meuller, Portfolio Construction with Alternatives, Toronto, June 4, 2015 30
Portfolio Examples: Max Sharpe & 1/N Data: Credit Suisse Hedge Fund Indices and MSCI World Equity Index Max Sharpe uses 60-month in-sample window Betas are rolling 60-month out-of-sample estimates What are the conditional and unconditional equity exposures of these portfolio strategies? Bobey & Meuller, Portfolio Construction with Alternatives, Toronto, June 4, 2015 31
Take-away Strategy Level A ti e strategies are ore orrelated tha traditio al metrics suggest, and Especially so during the worst possible time Portfolio Level Portfolio opti izatio that ig ores the strateg le el empirical stylized facts (Slide 2) will import these risks to the portfolio level Main take-away Traditional portfolio construction does not diversify across the a ti e strategies o plete risk spe tru ; Using traditional correlation metrics is one of (if not) the biggest model risk associated with active strategy portfolio construction Bobey & Meuller, Portfolio Construction with Alternatives, Toronto, June 4, 2015 32
One Solution: Reframe and Simplify Problem Portfolio construction with alternatives is Less about diversification across asset classes or traditional risk factors (i.e. cross-sectional style or industry exposures), and More about diversification across the empirical stylized facts outlined on Slide 2 Both perspectives contain their own sources of model risk But the second perspective is at least pointing in the direction of the core of the problem Bobey & Meuller, Portfolio Construction with Alternatives, Toronto, June 4, 2015 33
Example: Conditional Equity Exposure CS Hedge Fund Index and Sub-strategy Indices Scatter Plots to MSCI World Equity Presenter Name, Presentation Title, Toronto, Date 34
The ten sub-strategy pay-off profiles to global equity can be defined by four characteristic profiles Bobey & Meuller, Portfolio Construction with Alternatives, Toronto, June 4, 2015 35
Portfolio Construction by Pay-off Profiles The portfolio construction problem is now framed in by defining a target pay-off profile In the context of global equity, this means defining target upside and downside equity beta exposure Diversification examples Tactical diversifies Relative Value Defensive diversifies Directional The framework above allows Bottom-up active manager selection to be seamlessly integrated with Top-down strategic allocation and alternative beta or liquid replication allocation Bobey & Meuller, Portfolio Construction with Alternatives, Toronto, June 4, 2015 36
Does Convexity Pay-off? What are trade-offs of targeting a positively convex profile? Compared to traditional portfolio strategies, these portfolios tend to have Lower Sharpe ratios, but Higher tail-risk adjusted returns and Treynor ratios, lower lagged beta exposure, and more stable intertemporal beta exposure Bo e, B., Li, L. a d Meuller, L. 5 Does Co e it Pa -off? Portfolio Co stru tio ith Alter ati es, Worki g paper. Bobey & Meuller, Portfolio Construction with Alternatives, Toronto, June 4, 2015 37
Closing Alternative investment strategies tend to exhibit empirical characteristics that significantly deviate from the assumptions of traditional portfolio construction approaches Implication is that the traditional concept of diversification across assets or factors is very incomplete in this setting Ignoring these stylized facts is one of (if not) the biggest model risk associated with portfolio construction with alternatives One solution: use strategy pay-off profiles as building blocks to attain a target pay-off profile Bobey & Meuller, Portfolio Construction with Alternatives, Toronto, June 4, 2015 38