Benchmark change, composite exclusions and a board that wants to extend

The 2.1% is not established

• The benchmark changed mid-year. For nine months of the year the manager was measured against a benchmark that was not in force. The like-for-like comparison is a blend: three months against the old benchmark, nine against the new, or full-year against each, giving +2.1% against the new and −0.8% against the old. Both numbers must be shown.
• The new benchmark is structurally different: it moved from a 100% UK, unhedged mix to a global mix with 50% currency hedging and 10% cash. Much of the apparent excess return is an asset allocation effect - the trustees' own policy decision - not manager skill. If the manager had no discretion over the strategic mix, crediting them with allocation is wrong.
• The hedging decision was introduced at the same time. Sterling's move over the year will dominate a 45% global equity weight; the currency effect must be isolated before any skill conclusion.

Attribution the trustees should require

1. A Brinson-style decomposition - asset allocation versus stock/security selection versus interaction - run separately for the pre-change and post-change periods.
2. A currency attribution showing hedged and unhedged contributions, and the cost of the hedge.
3. Returns against both the old and the new benchmark, and against a blended benchmark for the transition period, with the benchmark in force at each date clearly stated.
4. Gross and net of fees, and the effect of the manager's cash holding against a 10% cash benchmark weight.
5. Money-weighted returns alongside time-weighted returns, since the scheme has cash flows and the trustees care about pounds, not percentages.
6. A rolling three-year number against the policy benchmark in force at each date, rather than one number against a benchmark chosen after the fact.

Governance and ethical issues

• Removing two accounts worth £38m from the composite in the third quarter is not consistent with a composite that claims to include all actual, fee-paying, discretionary portfolios managed to the strategy. Removing accounts because they underperformed makes the remaining record look better. Terminated portfolios must stay in the composite for the periods they were managed. The three-year +1.9% may therefore be misleading, and presenting it as compliant with a global standard without evidence of verification compounds the problem.
• The trustees' pack showing only the favourable benchmark and a single three-year number is a presentation choice that the trustees pay for; they are entitled to a complete picture.
• If the misstatement is material, the trustees should raise it formally with the manager's compliance function, ask for the composite definition, the account list and the basis of any compliance claim, and consider escalating to the regulator if the response is unsatisfactory.

Behavioural issues

• Anchoring on the three-year figure and recency bias from a good year.
• Framing: a benchmark change presented alongside a favourable comparison.
• Confirmation bias: the sub-committee wants to extend the mandate and is reading the pack selectively.
• Authority bias: deference to the manager's presentation and to a consultant who signed off the benchmark change.

Recommended response

1. Do not extend the mandate at this meeting. Ask for the full attribution and composite documentation first.
2. Pre-agree in the statement of investment principles the measure used for manager review - a rolling three-year excess return against the policy benchmark in force - so the goalposts cannot move again.
3. Appoint an independent performance measurer, or verify the manager's numbers, and require gross and net reporting.
4. Put the benchmark change itself on the agenda: was it a policy decision properly advised and documented, or a response to a year of underperformance?
5. Record the decision and the reasons in the minutes, including any dissent, and diarise a formal review in twelve months on a like-for-like basis.