Investment committees are widespread but often flawed. This monograph shows how structure and group dynamics distort decisions and offers a transparent, model-driven approach to improve accountability and outcomes.
Abstract
Investment committees (ICs) are ubiquitous in institutional asset management and arguably are overrated. This monograph treats an IC as a governance technology that maps dispersed beliefs into a single portfolio and shows how common design features can destroy value. In discussion-based committees, sequential deliberation and hierarchy generate (1) group shift — that is, final allocations that drift away from the arithmetic mean of member views, often in the direction of the salient narrative — and (2) judgement noise and weak accountability as a result of individual input being ignored in portfolio space and thus not being recorded to evaluate after the fact.
Support Research for the Global Investment Community. Your donation raises professional excellence and enables the foundation to continue its important mission to fund, publish, and distribute in-depth, high quality, independent research.
I propose an algorithmic alternative: anonymous portfolio-vector aggregation. Each member submits an explicit long–short portfolio vector, and vectors are scaled to a common ex ante risk level (equal influence in portfolio space) and then are averaged mechanically into a consensus portfolio, which is rescaled to the institution’s risk budget. The rule is transparent, limits chief investment officer capture by construction, and preserves diversification across opinions while tightening incentives through observable individual submissions.
Because field data rarely reveal both ex ante member portfolios and the counterfactual committee outcome under alternative rules, I use a large language model as a controlled narrative lab to generate member portfolios and stylised discussions across macro scenarios and to quantify governance-induced drift.