Considerations when selecting a Risk System
In our autumn 2012 edition we discussed the current state of risk management within the asset management industry and the concept of a risk 'toolbox' to aid investment decision making. In the subsequent follow up article in February 2013 we focussed on the importance of data and communication within the risk process. In this issue, we look at the challenges faced by asset owners and some of the considerations when selecting a risk system.
One of the objectives of pension funds is to have sufficient funds to pay towards the provision of a pension for its members, whether it is towards a 'pot' for defined contribution or to a more specific liability for a defined benefit scheme. The pension fund has to set an investment strategy to protect current payment streams and to provide for sufficient funding for future pension commitments. As such, there needs to be a growth element to the investment strategy balanced with a low risk element for protection against inflation and interest rate risks, to name but two risks. This presents the challenge of creating a process that provides the necessary metrics for the current total fund assets within the portfolio to be analysed in terms of risk exposures and to accurately model the future liabilities, allowing sufficient flexibility to adapt for changes in the investment strategy.
Pension funds also face a different challenge versus asset managers; namely that typically pension funds have to look across all their mandates as a single portfolio, whereas an asset manager can usually look at individual portfolios. This itself poses a data management challenge; collating the appropriate data from multiple sources in often multiple formats, amalgamating it, cleaning the data and filling in the gaps. Pension funds also have to calculate what their liabilities are for the next 30 years and beyond, taking into account specific assumptions and calculation complexities.
Pension fund risks
Many pension fund investment strategies and policy statements are centred around risks that are measured retrospectively on past performance measures. For example, using risk parameters such as historic ex post tracking errors; the classic looking in the rear view mirror. Even if such a historic risk parameter is breached, what does the pension fund do about it? Was the risk breach down to underlying risks in the investment strategy, or unusually volatile market events? Or was it because a volatile period is returning to 'normal'? Setting an investment strategy against past events in a complex and unpredictable financial environment is not sufficient to meet the needs of a pension fund. Risk is about understanding the negative outcomes in an uncertain future.
Liabilities are ever changing and influenced by specific changes in demographics, longevity etc, but are also impacted by other macro elements and the impacts are not linear. According to the Pension Regulator, when analysing DB schemes across the UK, a 0.1% fall in gilt yields may increase liabilities by 2%, but may only increase assets by 0.5% and a 0.1% rise in inflation, with nominal interest rates unchanged, can increase liabilities by nearly 1%. There are also impacts from changing interest rates and currency fluctuations with pension payments being made in sterling, any exposure to foreign exchange movements can be detrimental to the liability matching profile.
Setting asset managers a benchmark and then berating them for under-performing is far too easy when the absolute risk is partly from the market risk of the asset classes. In a falling market an asset manager may outperform their benchmark but an absolute loss does nothing to help the pension fund meet its liabilities. The overall pension fund benchmark should be considered in a liability relative way, where the benchmark is not the internal policy benchmark, but rather is a customised bond index that replicates the cash flow structure of the pension liabilities.
To calculate a liability benchmark and to calculate the present value of the liability stream is complex.
The investment strategy is liability driven with an overlay on the nominal amount of the pension fund's expected liabilities. This overlay can remove the unwanted risks of interest rate, inflation and currency movements. A risk system ought to be able to calculate the change in value of the liabilities and assets arising from a percentage change in interest rate, inflation rate and swap spreads at all maturities or key rate duration points along the yield curve. Again, the calculations can be complex and can vary from one system to another.
To replicate the liability expectations a benchmark portfolio is often created using zero coupon bonds. It is against this benchmark that the pension fund can be measured to assess how it is faring in terms of how the liabilities will be met. An overlay portfolio of swaps (inflation and interest rate), options, bonds and cash can be applied, in addition to currency derivatives. Further, some assets have hedging characteristics that can also be used to reduce risk, such as property and forestry.
Recognising that there are several moving parts to assessing the risks faced by pension funds, how are they measured and monitored? There are some established systems that can help, but there will be some element of specificity that is required in defining exactly which risks are to be monitored and the possible bespoke nature of the underlying calculations.
As always, there is an important data management challenge. Where will all the portfolio data come from and in what format? Who will source all the swap terms and conditions data on the swaps as there is no standard format within the industry right now? Who will manage this whole data sourcing, cleansing, gap filling and cross checking? Where will the market data come from? These are challenging questions for a pension fund to answer when resources and expertise are already stretched through the ever changing regulation landscape. Some system providers do offer a 'managed service' that is geared to taking this pressure off their client but at a cost. There is also the question of which data to use in terms of; can unaudited data be relied upon to accurately be used in risk measurement and management? There is a balance between using estimated information that is available daily and waiting for reconciled data to be available, perhaps only weekly/monthly.
But it is not just about reporting the risks as they stand today. It is becoming increasingly common place to 'stress test' the portfolio by simulating a market shock. For example, what would the liabilities and portfolio profile look like if we had another oil crisis tomorrow, a currency devaluation, another financial meltdown, or an interest rate hike? By doing these stress tests on a regular basis the pension fund can better understand how the assets would react and how the liabilities may change and thus reconcile the two. The risks that appear can then be hedged to future proof the pension fund as much as possible. Again, some systems have past market shocks in-built to use as scenarios and some have the ability to customise your own stress tests.
Once all the above is in place and all the appropriate risks are being measured and reported in a timely manner, who within the pension fund will then manage the risk? Will the Investment Committee be agile enough and meet frequently enough to make the appropriate investment decisions? Is there an in-house investment team to perform the 'what if' trades and does their risk system enable them to do so? Do they rely solely on an external party to do this, or is there a combination of the two?
Considerations when selecting a risk system
With all system selections, it is key to define the specific requirements to be able to pick the most applicable system. Some key features may include:
- Asset type coverage, including alternatives
- Ability to model liability streams and create a liability benchmark
- Data management and a managed service
- Clarity of metrics, e.g. which ex ante VaR calculation is more appropriate, parametric or historic simulation? How are liabilities discounted?
- Stress testing and 'what if' analysis
- Daily risk reporting
- Risk breach alerts
- Usability, drill down capability, data and calculation interrogation.
It is possible that no single system can meet all the requirements, so there may be a need to compromise on the risk system's capabilities if a single risk system is the preference. Alternatively, the requirements may only be a small part of a large risk system and so you pay for unwanted functionality. Or there is the best of breed approach, whereby multiple risk systems are selected. There are of course cost implications with all these approaches, but ISC has the subject matter expertise to assist clients in selecting the right approach given their requirements.