The first question that faces a pension fund with regard to performance measurement is why, and hence whether, it should be undertaken. In a UK context, where performance measurement is seen as the norm, this question would normally be seen as redundant. Here an extremely competitive fund management market, along with a trust law culture and the demanding requirements of pension law has led to extensive use of performance measurement. However, in less mature pension fund markets, or those without a trust ethos or extensive pension regulation, this question is still relevant.
In all facets of life, we want to be able to measure the performance of something that we have purchased, to determine whether it meets its purpose, our expectations, and whether an alternative might be more suitable. This applies whether the product is a widget or the investment services of the fund manager. Performance measurement belongs in the control domain of operating a pension fund. It enables pension funds to quantify performance, understand that performance, measure the risks that were taken, ensure compliance with the investment mandate and make informed decisions.
Once the decision is made to undertake performance measurement, the next decision a pension fund faces is who should undertake the task. A do-it-yourself approach is possible, particularly within a large in-house pensions department, but is normally impractical or undesirable, due to the complexity of the task, the cost-effectiveness or the lack of independence.
External providers include the investment manager themselves, custodians and specialist independent performance measurement providers. All have their own merits:
q An investment manager may offer it as part of a one-stop-shop service with no explicit price, but does not offer the comfort value of independence.
q Custodians have increasingly tried to leverage their relationship with pension funds to extend their service into performance measurement. They argue that they can provide a service of greater speed and consistency of reporting for the client than other providers. However, measurement cannot effectively be calculated until an account period is closed and audited, at which point all the data may efficiently be electronically transferred to another provider giving the custodian no advantage in respect of timing. With regard to consistency, this argument can hold true, but usually only in the case that the custodian operates globally for the pension fund, and hence is able to produce reports on all its investments in a consistent manner.
q Specialist independent measurers offer consistency across all funds that they measure, and, in countries such as the UK, where performance measurement has been used over an extensive period, with specialist measurers historically having the majority of the market share, they additionally offer access to a universe of results from other clients, against which the funds’ performance can be measured.
Product, accuracy, timeliness and price are usually the deciding factors in choosing the measurement provider. The costs of independent providers are usually similar within national markets, although there may be large differences in costs between different countries.
The next question facing a pension fund is how performance is defined, and hence what is measured. Within the context of a pension fund, the nominal return may have little relevance to the risks that the fund owner faces. Instead, measuring the performance relative to a benchmark may provide a more sensible view of the impact of a chosen investment decision.
A benchmark is a means of translating the level of risk that the owner of a fund faces to the investment manager. This is necessary as fund owners face different risks to investment managers. A fund owner has liability related risks, such as not being able to meet the liabilities, or some measure of the liabilities. A fund owner may also consider the risk of an opportunity cost arising by investing using the current investment manager rather than an alternative provider. Risks facing a fund manager include volatility of results, tracking error and business risk, which may lead to different behaviour, performance or risk controls to those wanted by the fund to meet its risks.
In measurement terms, the benchmark allows the performance of the investment manager to be measured against a baseline of the fund’s risks. Defining and refining appropriate benchmarks are decisions that fund owners will need to make both in implementing a manager and in measuring the performance. Indeed, evidence points to the fact that by laying down a strategic asset allocation, a pension fund has determined the bulk of its future returns. Performance measurement may also be used to judge the efficacy of a fund’s particular strategic asset allocation over any number of alternatives.
Germany is, or at least was, an example of a market which has typically focused purely on the nominal returns that a fund has produced, often without allowance for fees. It is essential for the German investment market that pension funds begin to measure their fund’s performance, with allowance for investment fees. Until this is achieved, not only will funds not be able to compare performance directly with other managers, but because of the charging structure, which typically has very low explicit fees with “hidden charges” incurred when stocks are bought and sold, the true cost of a particular manager will not be known. To compete on an international stage will require German investment managers to adopt GIPS, and this should be the driving force towards transparency in the German market.
In markets with a large usage of performance measurement, such as the UK, peer group benchmarks are still prevalent. In more developing markets in terms of measurement, such as Germany, the lack of data on peer managers means that at these early stages, bespoke benchmarks are the norm. Performance here has often been considered relative to the index return, which can still be misleading. For example in 1995, the German Dax equity index returned 7%, whereas the median manager, as measured by WM, returned 8%. Those funds that achieved the index, and were satisfied with their results, might not have been so - had they known that they had underperformed more than 50% of other fund managers.
However, in Germany, it is expected that as measurement becomes more prevalent, a universe of results will lead to, at least, parallel usage of peer group benchmarking. Ironically, in the UK, the Pensions Act 1995 has focused trustees, via the introduction of the minimum funding requirement and the statement of investment principles, more on their fund’s liability-based risks; the peer benchmark, which has been the leading benchmark for pension funds, is losing ground to bespoke benchmarks.
Pension funds need to decide how to interpret the results of performance measurement. As well as having the performance quantified, the pension fund should ensure that the manager is following the mandate that was given, particularly with regard to style and risk, and should understand the reasons for under- or over-performance. Performance attribution facilitates this understanding.
It is important to note that different measurement providers may produce different answers, with differences of 0.5% not being significant or unusual. Inaccuracy of results generally reflects inadequacy of the data, with cashflow timings being the biggest source of error. The UK’s two leading specialist independent performance measurers – WM and CAPS – until a few years ago differed in their calculations of the total return on the All-Share Index because of cashflow timing differences. Although their approaches were subsequently harmonised, they differed by 30 basis points on the total return of the World ex-UK Index in 1997. CAPS calculated this as 19%, and WM as 19.3%. Recognising these margins is important, as decisions are often made based on small differences in returns.
In interpreting, and acting on the results of performance measurement, it is critical to determine the timeframe over which results are assessed, and decisions will be made. The timeframe should be long enough to remove any random elements from the results, but short enough to allow funds to make decisions about results which are still relevant to the current (and presumably, future) capabilities of the investment manager. In the UK, a typical timeframe over which to consider a manager’s performance is three years. This has fitted in neatly with the control cycle of pension funds (for example, formal valuations).
Faced with poor performance, a fund will need to make a decision. Performance measurement facilitates the making of informed decisions. Changing managers is expensive, and requires an informed decision based on accurate data. Not changing managers can be more expensive, and still represents a decision.
Nigel Cresswell is a consultant and actuary with IPC, the international practice of Dr Heismann in Wiesbaden