Going fund-specific or the peer-group route
Within the European context, the extent to which fund-specific or peer-group benchmarks for investment portfolios are used will vary from country. One of the determinants has to be the state of development of the market. In the UK, for example, the peer group has long held sway, though the trend recently has been to more fund-specific benchmarks.
International consultant Watson Wyatt recently published its Global Asset Study (Asset Book 1998), in which it points out that, where there are large performance universes “the peer group becomes the dominant form of benchmark”. In the study, this is attributed to the lower level of responsibility that devolves on the plan sponsor and fund owners with peer-group benchmarks, while fund-specific benchmarks require taking on a greater level of responsibility.
In examining at first, fund-specific benchmarks, the study looks at optimisation, which it describes as asset allocations with the largest level of returns for any level of risk. This is measured by using volatility, with certain caveats. “Optimisation involves identifying portfolios that are optimal for minimising volatility at any level of return,” it says. This means looking at assumptions about returns, volatilities and cross correlations. The consultant argues that, in looking at equity markets, the return assumptions should be the same for each major market, as volatilities are similar. International investors are exerting influences in these markets where local investors used to dominate.
For future bond returns the bid yield as at December 31 1998 has been taken for each country, less an inflation element. Then the consultant took “a market capitalisation-weighted average of these estimates and used this as the bond return assumption for all markets”. The resulting estimate is 2% per annum real return for bonds.
For equity returns, the study uses the equity dividend yield plus forecast real GDP growth, again with the weighted average of these estimates. The final equity figure is 4% per annum real return.
Table 1 looks at return assumptions on a country-specific basis.
In looking at volatilities, the study comments that historical volatilities and correlations are more reliable than future estimates of returns, so data from the past 20 years are used.
The portfolios outlined below are those that minimise volatility for any level of return from that country's perspective. “Such portfolios are termed efficient and the set of portfolios identified in this way is referred to as the efficient frontier.” The figures on the facing page show the data relating to six European countries included in the global analysis carried out by Watson Wyatt in the study.
The charts show the different asset allocations on the ‘efficient frontier’, moving from a minimum risk portfolio to those with a range of expected returns (A to E) moving up from 2.4% real per annum to 4% in 0.4% steps and labelled as follows: Those in the tenth percentile are termed ‘best case’, in the 50th percentile (‘most likely’), 75th percentile (‘lower quartile’), 90th percentile (‘worst case’).
The study looks secondly at peer-group benchmarks, where it says the implicit aim is outperformance of the “average return from a performance universe of similar funds”. In such a scenario, the average asset allocation in a market is the benchmark. Table 2 shows the average of peer group portfolios.
Then using the return and risks analysis for the fund-specific benchmarks (Table 3) the expected returns and volatilities of country average portfolios are calculated. The combinations of returns and volatilities vary from the efficient frontiers in the fund-specific analysis. The study describes the sub-optimality as: “the shortfall in expected returns that could be achieved by an appropriate change of portfolio mix which leaves volatility unchanged”.
Though the expected shortfall by market for European countries due to the sub-optimality of peer group or country average portfolios is small: being only 0.1% pa in the cases of France, Germany, Netherlands and Switzerland, while in the case of the UK it was zero, but 0.2% for Ireland, making this country look the least efficient because of the low domestic equity allocation and the high return expectation due for this class.
The study says the particular choice of bond/equity mix is a result of risk aversion and the time horizon chosen, for both of which there is no objective measure. But the study as a guide creates a “time horizon measure by assessing the holding period required for the ratio of the real return to volatility to exceed one. This is the period required for the probability of a negative real return to decline to a one in six chance”.
The holding periods implied range from Ireland at 17.3 years to the UK 16, France 11.1, the Netherlands 10.6, Germany 8.6 and Switzerland 7.3. The comment made in the study is that some countries may be investing with a “too short focus, perhaps influenced by the need to meet short- term regulatory requirements”.
The upshot could be that they are paying too much for pension provision.
The cost of providing 100 units of pension benefit in relevant local currency in 15 years time is given in the study. This puts the UK lowest at 60 units, with Irelandat 63, the Netherlands 66, Switzerland 70, France 71 and Germany 72.
The study concludes that there is some evidence that peer group benchmarks may not be as efficient as investors would like.