Opalesque Industry Update - MSCI survey debunks popular myths about equity risk models|
MSCI Inc., a leading provider of investment decision support tools worldwide, announced today the results of a survey on best practices in risk modeling and portfolio construction, which debunks popular myths about the use of equity risk models in portfolio management.
Over 600 portfolio managers and other institutional investors globally responded to the survey, providing a practical assessment reflecting real beliefs and approaches to using models when building portfolios:
• ‘78% said that the quant meltdown was not caused by investors using the same model.’
• ‘88% said that not all equity risk models are the same and agree that individual providers have a distinct approach to model construction.’
• ‘58% said that equity risk models are useful when a financial crisis breaks and models are a key portfolio construction tool before, during and after a financial crisis.’
• ‘94% said that data quality is essential and important and without the best quality data, their portfolio construction process would not be successful.’
MSCI launched the Myths and Models Survey to attempt to understand how portfolio managers and other institutional investors view equity risk models within their portfolio construction process following the recent financial crisis. As the financial markets continue to be turbulent, it is important for industry professionals to understand what lessons can be learned from the past as we move into the future.
The majority of respondents (58%) who disagreed that equity risk models do not work when a financial crisis breaks believe instead that models are a key portfolio construction tool before, during and after a financial crisis, and can be used to analyze short and long-term volatility, as well as tail risk. A further 60% of respondents prefer a customized approach to the portfolio construction process, integrating a variety of country/regional and short-/long-term equity risk models supplied by an experienced third party provider into a single framework instead of building their own equity risk model.
Kurt Winkelmann, MSCI Managing Director and Head of Research at MSCI, said, “The MSCI survey results expose that the majority of those polled believe equity risk models are important and play a vital role in the clients’ day-to-day investment process. It’s not just the risk numbers, model output, data and factor structure, but the interpretation of these numbers within the context of the market environment that drives successful investment decision making.”
A majority (78%) of those polled said that the quant meltdown was not caused by investors using the same risk model. Rather they said risk models were used in different ways even if they produced the same ‘answers’ during this period, and that the interaction between alpha and risk models can yield different results. Feedback comments as to what those polled did think caused the meltdown included:
“When I saw my portfolio dramatically underperforming one day and found that the underperformance was totally consistent whichever way I sliced and diced (all factors, all sectors, all countries, etc.) then it was impossible to avoid the conclusion that one or many portfolios were being liquidated out there that had an uncanny similarity with my own. I believe both alpha and risk models contributed to the meltdown, but that I would say the alpha model was more significant.”
“We may or may not have used the same models, but we got to the same ‘answer’ (positions). When liquidation of these same positions started, it was a snowball effect. I would say alpha was more the ‘cause.’ A risk model just keeps you out of bets you don’t intend to take. If everyone used the same alpha model, risk model and optimizer, you’d get the same positions; but any of several competent approaches would get you close, and that’s probably what we observed.”
Of the institutional investors polled, 88% said that not all equity risk models were the same, and agreed that individual providers each have a distinct approach to model construction. Over half of the respondents said that responsiveness and a well designed structure that provided insight into sources of risk were the top two attributes that they look for in an equity model.
The overwhelming majority (94%) of institutional investors polled agreed that data quality was essential and important in the portfolio construction process and that without the best quality data, they would not be successful. MSCI takes the view that good quality data is a cornerstone of reliable and effective equity risk models.
The majority of institutional investors polled (82%) were aware of the sampling error that occurs when using models but were divided on the extent to which sample covariance matrices tend to over- or under-estimate risk. These findings highlight the biases that clearly exist in risk forecasts for optimized portfolios. MSCI believes that if left unchecked, this impact is significant. MSCI has developed the Eigenfactor Risk Adjustment that adjusts the covariance matrix and removes biases in risk forecasts to correct for this sampling error.
Mr Winkelmann said, “Investment insight, data quality and good models combined are essential in the portfolio construction process. The survey shows the extent to which investment managers, CEOs, Managing Directors, portfolio managers, quantitative analysts, and risk managers are incorporating multiple equity risk models to work in parallel within their investment processes to help them gain a better perspective and a more accurate view of risk in their portfolios.”
The MSCI Survey launched on July 22, 2011 and closed August 19, 2011.