Index Funds vs Active Funds: What the Data Says

• Sharpe's arithmetic (1991): before costs, the average active dollar must match the average passive dollar. After costs, it must lose. This is a proof, not a study.
• SPIVA data agrees: roughly 89.5% of US large-cap active funds underperformed over 15 years, ~92% over 20.
• Outperformance does not persist. Top-quartile funds rarely stay top-quartile.
• Fees are the one reliably predictive variable in fund selection, and they are known in advance.
• The passive case is strongest in efficient markets. It is genuinely weaker in less-covered corners — a caveat worth stating honestly.

The argument that needs no data

In 1991, William Sharpe — who had already won a Nobel Prize for the capital asset pricing model — published a short piece in the Financial Analysts Journal called The Arithmetic of Active Management. The argument runs like this.

Divide every investor in a market into two groups: those who hold the market in proportion (passive) and everyone else (active). Together they own all the shares, by definition — there is no third party holding the remainder.

Therefore the combined return of active and passive investors, before costs, must equal the market return. And since the passive group by construction earns the market return before costs, the active group as a whole must also earn the market return before costs.

Now add costs. Active management involves research, higher management fees, more trading and greater tax drag. Passive management involves very little of any of these. So after costs, the average actively managed dollar must underperform the average passively managed dollar. Necessarily. Every year. In every market.

What makes this powerful is what it does not assume. It says nothing about whether managers are skilled, whether markets are efficient, whether the era is bullish or bearish. It is closer to double-entry bookkeeping than to a theory. Active management is a zero-sum game before costs and a negative-sum game after them — one manager can only beat the average by taking that excess from another.

Sharpe's conclusion is worth stating in full, because it is often softened: properly measured, the average actively managed dollar must underperform the average passively managed dollar, net of costs — and this is a matter of arithmetic, not empirical evidence.

What the data shows anyway

Since 2002, S&P Dow Jones Indices has published the SPIVA Scorecard, comparing actively managed funds against the S&P benchmarks appropriate to their category. It is the most widely cited evidence in this debate, partly because it takes seriously two problems that flatter active funds in naive comparisons.

The first is survivorship bias. Funds that perform badly are quietly merged or closed, so a snapshot of surviving funds systematically over-states past performance. SPIVA tracks the original universe, including funds that no longer exist — and the attrition is substantial over long windows.

The second is style drift: comparing a fund to a benchmark that does not reflect what it actually holds. SPIVA compares like with like.

The results, for US large-cap equity funds against the S&P 500:

Period	Share underperforming
2025 (single year)	~79%
15 years	~89.5%
20 years	~92%

The one-year figure moves around a great deal — there are years when a majority of active managers beat the index, usually when market breadth is unusual. The long-horizon figures barely move at all. After 15 years, the odds of having picked a large-cap fund that beat the index are roughly one in ten, and that is before you consider whether you would have held it through the drawdowns along the way.

SPIVA's companion Persistence Scorecard closes the obvious loophole. If a tenth of managers genuinely outperform, could you identify them from their record? The persistence data says largely no: funds ranked in the top quartile in one period show little tendency to remain there, with rates frequently no better than random. This is the finding that makes the standard disclaimer — past performance is not a guide to future results — an accurate description rather than legal throat-clearing.

Where the difference comes from

Three costs, in descending order of visibility.

The expense ratio is the published annual charge. Broad index funds commonly sit at 0.03–0.20%; active equity funds typically at 0.5–1.0%. A persistent gap of 0.7% a year does not sound decisive, but over 30 years a portfolio compounding at 7% grows to 7.61× its starting value while one at 6.3% grows to about 6.26× — roughly 18% less final wealth, from the fee alone.

Trading costs are not in the expense ratio. Active funds turn over their holdings far more, and each trade incurs commissions and bid-ask spreads, and moves the market slightly against a large fund. These are real and largely invisible.

Tax drag applies in taxable accounts. High turnover realises gains that would otherwise stay deferred. Index funds trade rarely and, in ETF form, benefit from structural mechanisms that limit distributions. In a taxable account this can be worth more than the fee difference.

The honest counter-arguments

A one-sided case is a weak case, and there are real ones.

Market efficiency varies. The passive argument is strongest where it was tested most: US large-cap equities, the most analysed asset class in history, where finding mispricing means beating thousands of well-resourced professionals looking at the same filings. In small-cap, frontier and emerging markets, distressed debt, and parts of fixed income, coverage is thinner and dispersion is wider. Historically a larger share of active managers has added value there. Even so, the median fund often still trails after fees, and picking winners ahead of time remains unsolved.

Sharpe's arithmetic is about the average. It does not say no one can outperform — it says outperformance must be funded by someone else's underperformance. Skilled managers can exist and demonstrably do; the difficulty is identifying them in advance rather than in a backtest.

SPIVA's methodology has been challenged. Critics have argued about equal-weighting funds rather than assets, and about benchmark and category assignment, and at least one study has argued the scorecard overstates underperformance. These critiques narrow the gap in some categories. None of them reverses the long-horizon conclusion, and none of them touches Sharpe's arithmetic.

Index concentration is a genuine risk. A cap-weighted index owns more of whatever has already risen. The largest handful of US companies now make up an unusually large share of the S&P 500, so an "index" investor holds a considerably more concentrated portfolio than they may realise. That is a real exposure, and it is an argument for broader diversification across geographies and asset classes — not an argument for paying more.

"Closet indexing" muddies the comparison. Many nominally active funds hold something very close to their benchmark while charging active fees. They are structurally guaranteed to lose by roughly their fee. Their presence drags down the average active result while saying little about genuine active management.

What this implies in practice

The conclusion is less "index funds always win" and more a shift in what the decision is actually about.

You cannot reliably predict returns. You can predict costs. Fees are printed on the document and are among the very few variables with demonstrated predictive power for relative performance. Choosing on cost is acting on the strongest available information.

Diversification is separate from the active/passive question. An index fund tracking one country's large companies is cheap but not broadly diversified. Spreading across geographies, company sizes and asset classes is a different decision, and arguably a more consequential one.

Behaviour outweighs selection. The largest destroyer of real-world returns is not fund choice — it is buying after rises and selling after falls. Studies of investor returns versus fund returns consistently find a gap driven by timing. A boring portfolio you keep beats an optimal one you abandon in a crash.

If you use active management, do it deliberately. With a clear reason it can add value in a specific inefficient segment, at a defensible cost, and with an understanding that a decade of underperformance is a normal outcome rather than a signal to switch. What rarely works is buying last year's top-performing fund, which the persistence data says is close to noise.

Further reading: The SPIVA Scorecards are published free by S&P Dow Jones Indices, alongside the Persistence Scorecard. Sharpe's The Arithmetic of Active Management (1991) is three pages and freely readable.

This article is general information, not investment advice. Fund availability, tax treatment and regulation differ by country, and all investments can lose value.