Every holding of a plan was examined, and inserted into one of fifteen broad asset class strategies. The qualitative mandate of individual managers was not used (i.e., 100% of a stated large cap growth equity manager would not be mapped into a large cap growth proxy). Rather, each security was classified based on a series of rules to determine actual market exposure to one of the fifteen market index factors used in the model.
The fifteen asset classes were chosen to broadly represent investment vehicles worldwide. Plan exposure to each of these asset classes and sub asset classes was achieved by applying a series of rules on every holding to best determine the nature and type of security. For example, domestic equity issues were exposed to the Barra factors to determine allocations to large/small, value/growth orientations. For domestic fixed income effective duration was used to appropriately assign long bonds. Moody's quality ratings were used to determine exposures to investment grade and high yield bonds. For non US securities, country of incorporation was examined to determine whether the security needed a developed market proxy or an emerging one. In the case of commingled and mutual funds, where individual security data was not available, a returns-based style analysis was performed, and the historical return exposures used as proxies for actual assets held.
The fifteen asset broad asset categories and their corresponding market proxies are:
|Large Cap Growth||Russell 1000 Growth|
|Large Cap Value||Russell 1000 Value|
|Small CAP Growth||Russell 2000 Growth|
|Small Cap Value||Russell 2000 Value|
|Core Bonds||Lehman Aggregate|
|High Yield||Merrill Lynch High Yield|
|Long Bonds||Merrill Lynch Long Govt|
|Developed Equity Ex Japan||MSCI Ex Japan|
|Japan Equity||MSCI Japan|
|Emerging Equity||MSCI Emerging Free|
|Developed Bonds||Salomon World Govt. Bond Index Ex US|
|Emerging Bonds||JP Morgan Emerging Debt|
|Cash||90 Day T bill|
|Venture Cap||S&P 600|
The actual rules that were employed within the broader asset classes (Equity, Fixed Income, Venture Capital, Real Estate etc.) to determine style specific buckets are outlined below.
Domestic/Developed Market/ Emerging Market:
Country of incorporation of the security
Barra Size exposure > -1.80 indicates Large Capitalization Company
Barra Value > -0.58142 indicates a value orientation
Moody Quality < Baa3 indicates High Yield
Long Duration Bonds:
Effective duration > 6.5 indicates Long Bond
Parametric VaR Computation
The parametric VaR approach rests on the presumption that individual security positions can be mapped to a simple set of exposures each of which is affected by one risk factor, in this case an appropriate market proxy. Once a plan's allocation to the 15 buckets was determined, a parametric VaR was computed using a historical time series of returns to generate an estimate of the variance—covariance matrix of these market index factors. The idea underlying this approach is rooted in modern portfolio theory (MPT). Under MPT the expected portfolio volatility can be described as a function of the volatility of each individual security in the portfolio and the correlations between those securities. Because security prices are not perfectly correlated the total risk at the portfolio level is less than the sum of the risks of its component securities. This portfolio volatility can be described by the following equation.
Because we assume a normal distribution of portfolio values the portfolio volatility or standard deviation determines the VaR, which is simply the number of standard deviations that correspond to a given probability (E.g. 1.645 standard deviations to obtain a 95% confidence, a 1 in 20 occurrence)
While the current methodology provides an excellent broad-brush survey of the risks of institutional investors in the US today, additional information will be useful. Planned in the near future is inclusion of tracking error factors on a manager level basis to capture the risk of managers under performing their benchmark. Also, the inclusion of VaR relative to funded status is planned.
Calculating Your Exposure:
If you are a member of the Northern Trust Performance Universe, please contact Paul d'Ouville of the Risk Services group at 312-557-1716, or your Relationship Manager to determine the risk level of your plan within the universe.
Paul is a Senior Vice President at Northern Trust, Chicago. Paul leads the product management and development activities for Northern Trust’s Corporate and Institutional business. The broad range of products and capabilities support Northern Trusts’ strategic goals globally.
Paul leverages his experience in operations, technology, risk management and client service to ensure Northern Trusts clients optimally benefit from the solutions his team brings to market.
Prior to taking this leadership role, Paul headed the Investment Risk and Analytical Services division. This solutions oriented organization was focused on assisting clients with their risk management responsibilities.
Many initiatives Paul has participated in throughout the organization have brought new capabilities to clients or optimized the organizational efficiency of Northern Trust. These initiatives included the development of simulation methodologies for the bank's mortgage and derivative portfolios. Implementation of the initial Value at Risk model utilized by all foreign exchange trading desks. Regionalization of the IRAS division to support clients locally on a global scale.