Performance Measurement in Finance

Simulation as a means of portfolio performance evaluation 145

new instruments, such as derivative instruments, exchange-traded funds and

emerging markets securities.

• Data, tools and analysis techniques available to investment managers are

continuously evolving, placing inevitable pressure on investment managers

to upgrade and modify portfolio construction, analysis and monitoring tech-

niques.

• Regulatory and tax environments are subject to continuous change, forcing

investment managers to modify their approaches accordingly.

• Maintaining a reliable database of portfolio performance for the purpose of

multiple period portfolio comparisons for even a moderately comprehensive

range of investments is extremely difficult.

Portfolio return analyses therefore can usually span only a relatively small

number of periods, so that the results obtained nearly always fail to meet any

test of statistical significance or reveal persistent performance patterns.

The uncertainties implicit in traditional return and attribution analysis leave

plenty of room for speculation by investors that persistent outperformance

of benchmarks is due principally to luck or a particularly ‘easy’ mandate

specification; and conversely claims by investment managers that persistently

poor performance is the result of very tight or conflicting constraints. These

are claims that are difficult to reliably support or refute using conventional

return and attribution analysis.

Traditional attribution analysis cannot be applied to funds that do not reg-

ularly offer transparency of holdings to investors. Evaluation of such funds

is limited to comparison with their stated benchmarks or with funds with

similar stated investment objectives – comparisons of which may be of limi-

ted validity because of differences in objectives, focus, investment universe

or constraints, and because there are often only a small number of similar

portfolios available for comparison.

Conventional return and attribution analysis does not facilitate ‘what-if’

analyses, aimed at exploring the performance effect of changing various port-

folio selection parameters, for example altering the scope of the investment

universe and testing changes in hedging policy.

Moreover, these problems are all equally acute at the level of asset

allocation for balanced portfolios and security selection for specialist sector

portfolios.

6.2 OBJECTIVES OF SIMULATIONS

The objectives are:

• To distinguish portfolio outcomes that are due to chance from those

resulting from deliberate investment choices and systemic biases in