Deciding Whether to Terminate an Active Manager
An investment committee is reviewing a UK equity income manager after a difficult period. The mandate is benchmarked to the FTSE All-Share. Over the three years to 31 March, the manager returned 7.8% annualised net of fees against 8.4% for the benchmark. The fund's realised tracking error over the period was 5.2%, and the annual management charge is 0.75%. Over five years the manager returned 9.1% annualised versus 8.9% for the benchmark. The lead manager changed the stock-selection process 18 months ago, reducing the number of holdings from 65 to 42 and increasing active share from 55% to 72%. The committee's policy says managers are placed on watch after two years of underperformance and terminated after three, but the policy was written before the process change. The committee meets next week.
1. Assess whether the three-year underperformance is statistically meaningful.
Read the worked answer
Annualised active return is 7.8% − 8.4% = −0.6%. With a tracking error of 5.2%, the information ratio is −0.6/5.2 ≈ −0.12. To judge whether a three-year average is meaningful, approximate the standard error of the three-year mean as 5.2/√3 ≈ 3.0%. The t-statistic is therefore about −0.6/3.0 = −0.2. That is far from conventional significance: a manager with true zero skill would routinely produce a three-year result this far below benchmark. The evidence is consistent with noise, not proof of a broken process.
The fee also matters. Net return is 7.8%; adding back the 0.75% charge gives a gross return of about 8.55%, slightly above the 8.4% benchmark. So the net shortfall is largely, perhaps entirely, explained by fees rather than by poor security selection. That does not excuse underperformance, but it changes the diagnosis from 'the manager cannot pick stocks' to 'the gross alpha is too small to cover the fee'.
2. How should the five-year record and the process change affect the decision?
Read the worked answer
The five-year record is 9.1% versus 8.9%, a small positive annualised active return. Combined with the negative three-year number, this suggests no consistent evidence of skill or failure. The process change 18 months ago is more important: it altered the portfolio's risk profile by cutting holdings and raising active share. That means the recent tracking error and the recent underperformance may reflect a new, more concentrated strategy, not the strategy that produced the five-year record. Comparing the three-year number with the old policy is therefore misleading.
The committee should ask for attribution over the post-change period: was underperformance from stock selection, sector allocation, or factor exposure? Was risk within the mandate? Did the manager follow the stated process? A policy rule written before a material process change should not be applied mechanically. The right response is a targeted review, not an automatic termination based on a backward-looking threshold.
3. Recommend an action and set out the criteria for the next review.
Read the worked answer
Retain the manager but place the mandate on formal watch rather than terminate immediately. The statistical case for termination is weak, the gross return is marginally ahead, and the process change needs time to be evaluated. However, the committee should not simply wait: it should require a written explanation of the post-change performance, holdings-based attribution, and confirmation that risk remains inside the mandate.
Set explicit review criteria: for example, review after a further 12 months; terminate if active return remains negative after fees and attribution shows the loss is from the new stock-selection process rather than a market-wide style effect; or terminate earlier if there is style drift, unexpected concentration, key-person departure, or a risk-limit breach. The committee should also record that the three-year result was not statistically significant and that the policy threshold needs updating for process changes. This is a judgement about evidence and process, not a mechanical rule.