Option B is correct. Empirical studies and theoretical models show that there is a proportional relationship between a portfolio manager's tracking error and portfolio return dispersion. Higher active risk (tracking error) leads to wider potential deviations in the specific weights held in different accounts, increasing dispersion.

Option A is incorrect. Dual-benchmark optimization is highly restrictive and typically leads to constrained portfolios that may not necessarily achieve higher average returns. Option C is incorrect. Dispersion is driven by both client-specific constraints (e.g., tax situations, specific restrictions, timing of cash flows) and manager-driven factors (e.g., trade execution delays across accounts, block trade allocation). Therefore, it is not purely client-driven. Option D is incorrect. Reducing dispersion to absolute zero is neither practical nor recommended. Clients possess individual risk tolerances, cash flow needs, and constraints, which inherently warrant some level of acceptable dispersion.