By now, it is well accepted that a pension fund should be viewed as a business with a balance sheet and income statement. The former reflects the market value of the investments that have been undertaken in the fund, the present value of the future pension liabilities and the balancing value of the 'surplus'. This latter account is equivalent to the shareholders' equity in the normal balance sheet. As in any business, the level and risk of the surplus account is a decision variable of management. The income statement relates to the costs of undertaking the investment and governance functions of the pension fund. The level of surplus impacts the likelihood of meeting the pension fund liabilities as they come due. While there are differences in the interpretation of the surplus for public and private pension funds, its importance is well understood by all pension funds.

The participants at the 1999 Risk Management Conference represented pension funds of varying sizes. Investment management methods and systems of risk measurement differed among the participants, with the distinguishing characteristic being the size of the fund and the associated economies of scale. For example, the large pension funds typically employed 'in-house' investment management with, in some cases, a small amount of the management allocated to outside portfolio managers. The smaller funds delegated fund management either to a single external manager or a set of external managers. The choice of manager(s) and related investment style were consistent with the philosophies of the pension funds with respect to their stated goals of risk exposure and expected return.

A similar distinction between large and small funds in the risk measurement and control process became clear in the lively and frank discussions during the conference. All participants agreed that risk assessment and management were important and should be undertaken on an ongoing basis. However, the approach to achieve these goals and the variables emphasized in the analyses differed among the funds represented. For example, although all participants agreed that the risk of the surplus was important, many chose to assess only the risk of their investment portfolios. The larger funds assessed risk in terms of surplus as well as the investment portfolios using costly and in some cases, complex systems that included Value at Risk (VaR). A specific and highly sophisticated application of risk assessment (VaR) was presented at the conference, and the importance and structure of various risk assessment systems was stressed. Those who spoke on this subject agreed that the cost was high for the systems and this realization led many smaller funds to confront 'sticker shock'. However, what was less well appreciated by some of these funds was that the size of the investment necessary to obtain a good risk measurement and control system when considered as a function of the fund size was not outrageous. Further, pension fund management, after identifying the risks it faces, must choose those they will control. This decision will influence the benefits and costs of risk management systems.

Many of the smaller funds described the risk assessment methodology used for their investment portfolios. The assessments referred to the possible portfolio rates of return outcomes based on a set of assumed scenarios - a simulation analysis. However, questions still remained as to how simulations were undertaken in the case of multiple managers, since it is the overall portfolio that is of interest and the variability of this portfolio is related to the variability of individual portfolios and the interrelationships of these portfolios' rates of return. To resolve this issue requires a more sophisticated approach and improved inputs.

One important risk measurement technique discussed in some depth at the conference was Value at Risk. All participants agreed that this approach was one of a number of risk measurement and control tools available and that reliance on only one may not be sufficient. VaR has similarities with the simulation techniques currently utilized by some funds. Its success depends upon the quality and representatives of the inputs used in the analysis. An interesting dynamic at the conference was the linkage of the VaR approach with high cost and complexity and hence its limited applicability to smaller funds. However, what appeared to be less well understood was that any simulation system that assesses the risk of the surplus and the overall portfolio risk would require inputs and technology that are expensive. Also, the most sophisticated systems are not needed for the risk analysis. The current techniques used by smaller funds may not be delivering the appropriate information for risk control, since the inputs and technology are limited. Another concern about the use of VaR related to the complexity of the system and the expected difficulty of communication to trustees. However, if risk control is important, an effective communications system needs to be developed. This is part of the cost of using the system.

The conference participants had an interesting discussion about methods through which smaller funds could obtain the necessary inputs at a lower cost. Agreement was quickly achieved that funds can not rely on fund managers to provide the needed information, since it would be very hard for the pension funds to audit the information. Changes in investments or in weights assigned to various asset classes could alter the risk of portfolios, but this information may not be provided in a timely way. This problem is similar to the issue of fund management style and assurances that style is being maintained. The issue of information provision becomes even more problematic when there are multiple managers. In this situation the correct assessment involves the complete portfolio and not the individual portfolios. Only by having the complete portfolio or the correlations of the rates of return on the portfolios associated with each manager can correct risk assessment/control be undertaken. Addressing this problem is similar to ensuring that the amalgamation of all the sub-portfolios does not result in the equivalent of a market index overall portfolio inadvertently generated at a high cost.

In meeting their responsibilities, pension funds need to assess the risk and level of surplus along with the variability of the investment portfolio. Technology is being developed that will permit smaller funds to use sophisticated methods of risk management. The problem in the use of any of these techniques, some of which are currently in use, is the generation of relevant inputs to the process. Pension funds will have to find cost-effective ways to obtain this data and convince themselves that the data they have reflects the composition of the entire portfolio of assets and liabilities. These issues were considered in depth at the conference, both in terms of presentations and discussions. For the sake of the fund sponsors and beneficiaries, it is hoped that pension funds address the challenge of improving risk measurement and control systems.

Paul Halpern is TSE Chair of Capital Markets and Director of the University of Toronto Capital Markets Institute.