Gail Moss finds out how pension funds can use benchmarking to reduce their operating costs
For many years, pension fund performance has hogged the limelight as an issue of fundamental importance. Investment return generally makes up the bulk of a pension fund’s income, and (in contrast to member contributions) is a variable quantity which can be maximised by choosing the right strategies.
Yet pension funds can also improve their financial position by controlling their costs. Consultancy firm Towers Watson estimates that the total cost of running a pension scheme has risen by 50% over the past five years and for large funds it now averages around 110bps, compared with 65bps in 2002.
An important tool in cost reduction is benchmarking, which allows funds to see where there might be scope for belt-tightening. “Benchmarking is a key tool in stopping the erosion of alpha,” says Ben Gunnee, European director, Mercer Sentinel. “But while pension funds are familiar with benchmarking an investment manager’s portfolio performance, the technique is still relatively underused beyond that.”
One reason for this is that benchmarking costs is not as easy as benchmarking performance: while some types of fees are measurable, there are other, hidden costs which are difficult to identify.
A basic ratio to benchmark is costs per member. According to the 2009 Pension Scheme Administration Survey from Capita Hartshead covering UK schemes, the average administration cost per member for schemes with under 2,000 members was £97 (€112) per annum for those using in-house administration services, and £101 per annum for those using external administration. At the other end of the scale, the use of third-party administrators appears to have paid off, with average annual administration costs for pension funds with more than 10,000 members being £46 for in-house and £39 for external administration.
However, broad-brush comparisons do not tell the whole story. According to John Simmonds, director of Toronto-based CEM Benchmarking, the key to efficient cost control is to understand what factors drive cost. CEM provides a service that analyses five factors - transaction volumes, economies of scale, cost environment (for instance, location), complexity and the service levels that the pension scheme provides. It then benchmarks schemes against their peers.
“Transaction volumes are the most important driver of costs,” says Simmonds. “Pension funds need to know if they are doing more or less work than their peers and decide what is necessary or unnecessary from a service-level perspective.”
High costs are not necessarily a bad thing. For instance, the scheme might be relatively sophisticated, or provide a high level of service. “It is fine to be a low-service, low-cost fund or a high-service, high-cost fund, as long as that is the clear objective of the trustees,” says Simmonds.
But when presented with a comparison of costs, it is usually possible for pension funds to get to the heart of why some specific functions are so expensive. And the principle is the same across Europe. “Essentially, the same things drive the cost of pension admin wherever you are, although there are different issues in each country,” says Simmonds. “What makes UK funds complex, say, is different from what makes Dutch funds complex. For instance, the Netherlands does not have contracting-out legislation. But the underlying principles are the same, and we benchmark pension funds against funds in the same country. However, a lot of funds are now interested in cross-border comparisons.”
One way for larger pension funds to offset costs is to convert their back offices from cost centres to profit centres, selling admin services to other funds. Two examples have been the UK’s Railpen and Siemens pension funds. But for those pension funds that cannot do this, there are many costs which can be pared down.
Fund management fees take the lion’s share of the total costs of running pension schemes, according to Towers Watson. It says pension funds should now be driving a harder bargain on fee structures, as they are too high for the value they offer. “The first step in getting fees back in line is for investors to understand what proportion of alpha their managers are taking in fees, even if they are doing a good job,” says Craig Baker, the firm’s global head of manager research. “This needs to be set at a reasonable level, and better structured to align interests.”
He says managers should always set hurdles, such as T-bills or a realistic fixed percentage reflecting the expected long-term beta exposure. They would then collect fees only on performance above this benchmark rate. The hurdle rate should also reflect a fund’s market exposure - for instance, a long/short equity fund with an average net exposure of 70% should base its hurdle rate on 70% equities and 30% cash.
Instead of calculating performance fees based on annual results, fees should be charged on a rolling three-year (or longer) basis. Baker says that base fees should also reflect actual costs, rather than the value of assets under management, because this encourages asset gathering, which can harm performance.
Other more routine charges, such as the fees charged by custodians, also offer scope for savings. “Pension funds are under the impression there is no difference between custodians, but there is,” says Gunnee. “And you can measure their performance to ensure they are not eroding alpha.”
However, custody fees can only be reduced for segregated clients, as investors using pooled funds will be charged for custodianship within the overall fee. One way of measuring custody fees is to use transaction cost analysis (TCA), a standard technique in the US.
TCA uses data from individual trades to measure not only brokers’ commission but also the market impact on the price caused by the trade, and also the spread - the market makers’ profit. “TCA shows whether the broker or investment manager did better or worse than the average,” says Gunnee.
A benchmark for custody fees could, for instance, be a service level with key performance indicators. Typically, these include the rate of interest paid by custodians on cash balances, and the rate of interest pension funds pay on overdrafts. Pension funds should compare them with what other custodians are paying or charging.
A further benchmark is the quality of foreign exchange executions. These are less transparent than share trades but some costs can be checked. For instance, if the custodian provides time-stamp data, the price achieved at the time of the transaction can be compared with market rates.
Timing is also important in terms of the frequency and speed of payments made by the custodian to the client. For example, how often does the custodian process income collections such as coupon payments. Are these sent to the client as soon as possible, or is there a delay? And how often does the custodian reclaim tax from other countries on behalf of clients? “Most clients don’t monitor this, but they could be losing 3bps here, 4bps there, and before they know it they’ve lost 20bps,” says Gunnee.
Besides benchmarking costs, pension funds can use other ad hoc measures to reduce their spend. For instance, they can ask their investment manager to rebate part of their commission.
They can also earn a modest return to help offset admin costs by offering services such as securities lending - the practice of lending equities, bonds or other assets in return for a fee to a third party. The borrower has to provide collateral - cash or securities - to at least cover the value of the securities being borrowed. “Although the credit crunch has made it more difficult for borrowers to provide collateral, a well-thought-out securities lending policy can be a low-risk investment offering some return,” says Gunnee.
However, Towers Watson recommends that pension funds do not embark on securities lending unless they fully understand the loan arrangements and risks. Where funds do offer securities lending, Towers Watson recommends increasing the collateral requirement where necessary, reviewing the list of borrowers and the indemnification structure, and changing the cash collateral re-investment guidelines. It also recommends using conservative non-cash collateral guidelines or very conservative re-investment of cash collateral as risk minimisation strategies.
Besides costs per member, another significant ratio for a pension scheme is the member-to-staff ratio - that is, the total number of its members divided by the number of staff. “This ratio is an indication of a scheme’s efficiency,” says Lorraine Harper, head of UK retirement consultancy services, Hewitt. “A ratio of 500 to one, say, might indicate a highly manual information services system. It could, however, be because volumes or complexity are high.”
By contrast, a figure of 3,000-plus to one could mean the scheme uses a high level of automation or call centres which deal with a large percentage of queries. This can lead to efficiency but could mean that members receive a relatively low standard of service.
In the long run benchmarking cannot measure everything about service performance. “The time it takes to get a report out can be benchmarked, but not the quality of the writing,” says Harper.