Introduction

This research provides the first large-sample evidence that decentralised voting is widespread within mutual fund families, challenging the prevailing academic view that these institutions vote as unified, centralised blocs. While much of the existing literature assumes that fund families operate as single decision-making units, this study finds that significant decentralisation in proxy voting has existed for years, predating the introduction of modern options like pass-through voting. Between 2006 and 2023, approximately 40% to 60% of US equity fund families exhibited evidence of decentralised voting, demonstrating that internal disagreement is a persistent feature of the industry.

Data and descriptive evidence

The study utilizes shareholder voting records from ISS Voting Analytics covering the 2006 to 2023 proxy seasons. The primary measure of decentralisation is within-family disagreement, defined as instances where funds within the same family cast non-unanimous votes on a given proposal. While the overall rate of disagreement is low at approximately 4.2%—largely because the vast majority of proposals are routine management items receiving near-unanimous support—this figure masks significant fragmentation on contentious issues.

Disagreement rises sharply to between 15% and 20% for proposals where the final vote outcome approaches the 50% majority threshold. Internal divergence is also notably higher for environmental and social (E&S) shareholder proposals, which see an 18.1% disagreement rate. Furthermore, external signals of contention, such as negative recommendations from proxy advisors like ISS or Glass Lewis, trigger greater scrutiny and divergence. For instance, disagreement increases from 3.2% when ISS recommends a vote in favour to 13.2% when it recommends a vote against management.

Determinants of decentralisation

A fund family’s decision to decentralise reflects a trade-off between leveraging specialised information and economising on stewardship costs. Families with a higher proportion of actively managed funds exhibit greater decentralisation, as active managers are more likely to possess firm-specific information. Conversely, “robo-voters”—institutional investors that closely follow proxy advisor recommendations—show significantly less internal disagreement, indicating limited stewardship capacity.

Organisational structure also plays a critical role. Families that employ external sub-advisors are 3.7 percentage points more likely to exhibit within-family disagreement compared to the unconditional mean. This supports the view that voting authority is often delegated alongside portfolio management responsibilities to those best positioned to evaluate proposal impacts on specific firms.

Consequences for governance

The research establishes that centralised voting is more effective for institutional monitoring than decentralised voting. Director elections serve as a key metric for this assessment; the impact of shareholder dissent on a director’s departure is strongest when a fund family presents a unified front. When voting families present a united message of disapproval, the probability of director turnover is markedly higher. Specifically, the average level of disagreement observed in practice diminishes the sensitivity of director turnover to shareholder opposition by approximately 15%.

This result is corroborated using Vanguard’s 2019 adoption of decentralised voting as a quasi-natural experiment. Following this policy shift, which moved authority from a central group to external managers, a dissenting vote against a director from Vanguard became significantly less likely to lead to that director’s departure. This suggests that fragmented voting dilutes the signal sent to corporate boards and weakens board accountability.

Consequences for fund investors

For fund investors, decentralisation appears to function as a product differentiation strategy rather than an informational advantage. Funds that deviate from their family’s voting stance are found to be more expensive, charging approximately 0.062% more per year in fees, which represents a 9% increase relative to the mean expense ratio. Despite these higher costs, deviating funds do not deliver superior performance. In fact, they exhibit negative gross alpha of approximately -0.74% annually.

Interestingly, while retail clients do not appear to react to these deviations, sophisticated institutional investors reward tailored voting strategies. Funds that deviate from their family stance attract significantly higher investor inflows when offered to institutional clients. This indicates that these investors value customised voting that aligns with their specific governance priorities, even when it does not result in higher financial returns.

Conclusion

The study concludes that the “one institution, one voice” assumption is often incorrect. Internal complexity and heterogeneity within fund families mean that their influence on corporate governance is more fragmented than commonly recognised. While decentralisation caters to the diverse preferences of institutional clients, it carries the significant trade-off of weakening the monitoring effectiveness of large shareholders. These findings provide timely evidence for the ongoing debate regarding pass-through voting initiatives, suggesting they may further undermine institutional monitoring effectiveness.