Project Alexandria

Active Fund Underperformance

Properties2
tagsdomain/finance, type/concept, level/intermediate, status/complete
aliasesActive vs Passive, Berk-Green, Mutual Fund Alpha Puzzle

11 min read

Active Fund Underperformance

The active fund underperformance puzzle is the empirical fact that the average actively-managed equity mutual fund delivers negative after-fee alpha relative to a passive benchmark, while households continue to allocate the majority of equity mutual-fund assets to active rather than passive products. The puzzle has two pieces that must be held together: a benchmark that says what after-fee returns should be, and an empirical result that says what they actually are.

The rational benchmark (Berk and Green, ). In a frictionless world where (i) fund managers possess genuine skill, (ii) there are decreasing returns to scale — a $ billion fund cannot deploy its alpha-generating strategy as efficiently as a $ million one — and (iii) investors are fully rational and Bayesian, flows chase past performance until each fund has been scaled up to the point where its after-fee alpha is exactly zero. Skilled managers still produce gross alpha, but they capture all of it in fees; the marginal investor is indifferent between the active fund and the passive benchmark. The Berk-Green equilibrium is the rational null hypothesis against which all active-fund evidence is measured. It does not predict that active managers are untalented — it predicts that talent accrues to managers, not investors.

The empirical result. The after-fee alpha of the average actively-managed equity fund is not zero, it is negative — on the order of to per year across the three classic studies:

  • Gruber () — first formally documents that the average active equity fund underperforms its benchmark after fees, using the full post- CRSP mutual-fund database.
  • French () — calculates the aggregate “cost of active investing” (fees, trading costs, and tax drag) and finds that active investors collectively pay about of aggregate market value per year relative to a passive benchmark.
  • Fama and French () — apply a bootstrap test to the cross-section of active-fund alphas and find that after fees, the distribution of -statistics is shifted left of the rational prediction: the top funds look roughly consistent with luck, and the bottom funds are worse than luck.

These findings violate Berk-Green: if the Berk-Green mechanism were running, average after-fee alpha would be zero, not -to-. So either the model’s assumptions fail or investors are systematically allocating to dominated products.

Channel decomposition (Del Guercio and Reuter, ). The underperformance is not uniform across active funds — it is concentrated in broker-sold funds, i.e. funds distributed through intermediaries (brokers, insurance agents, financial advisors) who receive commissions. Funds sold direct-to-investor (e.g. Vanguard, Fidelity Select, T. Rowe Price no-load funds) show roughly zero or slightly positive after-fee alpha, consistent with Berk-Green. This decomposition reframes the puzzle: it is not that households prefer losing money to active managers. It is that households who make independent choices behave roughly rationally, while households who delegate to a commissioned intermediary are sold into dominated products. The underperformance puzzle is therefore a distribution-channel problem, not a direct household-cognition problem.

Why the puzzle matters. Active funds hold more than half of U.S. equity mutual-fund assets. If Del Guercio-Reuter is right, the welfare loss from broker-channel misallocation is large — on the order of tens of billions of dollars per year of household wealth transferred to intermediaries. It also reshapes the policy debate: the remedy is not “nudge households to pick passively” but “change broker compensation structures” — which is the motivation for the SEC’s Regulation Best Interest and similar fiduciary rules.

Intuition

The Berk-Green model begins from an observation that most discussions of active management miss: genuine skill should be scarce, and scarcity should produce rents. A truly skilled stock-picker whose trades generate of alpha on a $ million portfolio will attract inflows, because rational investors see the track record and pile in. As assets grow, the strategy becomes harder to execute — prices move against the manager’s orders, liquidity dries up in small-cap names, decreasing returns to scale bite. At some fund size, the gross alpha has shrunk exactly to the level of fees, and any further inflow would push after-fee alpha negative, so flows stop. In equilibrium, every skilled manager has been scaled up to the point where investors are indifferent, and every unskilled manager has been allowed to die of underperformance. Talented managers end up rich; investors end up at zero alpha. Berk-Green is, in this sense, a rational story about why active funds don’t beat passive ones — but it is also a story that predicts zero underperformance, not negative underperformance.

The empirical data sit a full – percentage points below that zero line. Somebody is receiving dominated products — either because they don’t know better, or because someone else is choosing for them. Del Guercio and Reuter’s channel decomposition is the crucial piece: it shows the losers are disproportionately concentrated among investors who delegate the choice to a broker. The broker earns commissions on high-fee active funds but nothing (or little) on low-cost index funds, so the broker’s incentive is to recommend the active option regardless of its expected return. The household trusts the broker, does not separately verify, and ends up holding the dominated product. Direct-channel investors face no such conflict and their allocations do not underperform.

Properties

(Berk-Green Equilibrium) Flows absorb skill until after-fee alpha is zero.

Berk and Green () assume managers have persistent skill and costs are convex in fund size. A rational Bayesian investor observes past returns, updates her belief about skill, and allocates capital until the marginal fund offers the same expected return as the passive benchmark. At the fund level, the skilled manager’s gross alpha decays as assets grow because large trades are harder to execute; at the equilibrium fund size, gross alpha equals fees, so after-fee alpha is exactly zero. The model generates flow-performance sensitivity (“money flows to past winners”), persistence in gross returns (“skill is real”), and zero persistence in net returns (“rents go to managers, not investors”) — all three of which are broadly consistent with the data at the top of the active-fund distribution. The model is falsified at the average: the data show negative net alpha, not zero.

(Magnitude) Average after-fee alpha is negative – per year.

Gruber () documents that from –, the average active equity fund underperformed the market by about basis points per year after fees. French () widens the lens: combining fees, trading costs, and the tax inefficiency of actively managed funds, the aggregate “cost of active investing” is roughly basis points per year of aggregate market capitalization, or several tens of billions of dollars annually in the U.S. alone. Fama and French () bootstrap the cross-sectional distribution of alpha -statistics and find that after fees, the distribution is shifted systematically leftward of what pure luck would produce — there are fewer high-alpha funds than chance, and more low-alpha funds than chance. The three studies triangulate the same conclusion with different methods.

(Broker Channel Concentration) Del Guercio-Reuter decompose the underperformance by distribution channel.

Del Guercio and Reuter () classify the U.S. active-fund universe into broker-sold and direct-sold and compute alphas separately. Broker-sold funds — those distributed through intermediaries that collect front-load or ongoing commissions — account for essentially all of the documented underperformance; direct-sold funds (no-load funds marketed directly to end investors) earn roughly zero after-fee alpha, consistent with Berk-Green. The cleaner the channel, the closer the data get to the rational benchmark. This splits the puzzle into a solved piece (direct investors are fine) and an unsolved piece (broker-sold is dominated), and pushes the explanation toward incentives in the distribution channel rather than cognitive errors on the part of households.

(Fama-French Bootstrap) The left-shifted alpha distribution is hard to rationalize by luck.

Fama and French () ask: if managers had zero skill and generated only noise, how would the cross-section of alpha -statistics look? They simulate the counterfactual by bootstrapping residuals from a Carhart four-factor model and comparing the simulated distribution to the actual one. The right tail of the actual distribution is thinner than the bootstrap — there are fewer stellar performers than chance would produce, inconsistent with a large population of truly skilled managers — and the left tail is fatter — more dreadful performers than chance would produce, consistent with fees and transaction costs grinding down the returns of unskilled managers. The shape of the distribution suggests active managers are slightly worse than zero-sum after fees.

(Why Households Stay In Spite of It) The puzzle is distributional, not aggregate.

If investors knew active funds lost per year, they would all switch to index funds overnight, and the puzzle would disappear. They don’t — and the reason is not that investors are uniformly irrational. Direct-sold investors do reallocate toward passive: Vanguard’s market share has grown from under to over of U.S. equity mutual-fund assets over the past two decades as direct-channel investors learned the lesson. The holdouts are disproportionately broker-served households, who trust an intermediary whose compensation depends on recommending the high-fee option. In aggregate, active funds lose market share every year, but slowly — because inertia, existing commissioned relationships, and the complexity of the fund menu all delay adjustment.

Examples

The Berk-Green zero-alpha equilibrium

Suppose a manager has real skill: on $ million in assets, she generates gross alpha per year. Her fee is . At this size, after-fee alpha is , which is huge — rational investors see the track record and pile in. As assets grow to $ million, her trades start moving prices and gross alpha declines to . After-fee alpha is now . Flows continue until at $ billion in assets, gross alpha has decayed to and after-fee alpha is exactly . At this point the marginal investor is indifferent between her fund and the index, and flows stop. The manager captures 1\% \times \1\text{B} = $10$ million per year in fees — her rents from genuine skill. Investors earn index returns. This is the Berk-Green equilibrium: skill is real, skill is rewarded, but investors see none of it. Observed after-fee alpha should be zero, not negative.

See Also

Graph View

Backlinks