Excess return is an investment’s return minus a stated reference return over the same period. In asset pricing, the reference is commonly a risk-free rate or a cash proxy for it. In performance reporting, “excess” may instead mean return above a market benchmark. Naming the reference changes the investment question: did the portfolio beat cash, or did it beat the opportunity available in its market?
For a portfolio return and reference return , the arithmetic excess return is
If both inputs are percentages, report their difference in percentage points. A positive result means the portfolio beat that reference during the measured period. It does not establish that the return compensated adequately for risk.
One portfolio, two legitimate comparisons
Consider an original synthetic example with nominal USD total returns over one matched holding period. All distributions are reinvested; there are no fees, transaction costs, taxes or external cash flows. Each alternative starts with USD100.
| Investment | Period return | Ending wealth |
|---|---|---|
| Portfolio | 8% | USD108 |
| Cash reference | 3% | USD103 |
| Equity benchmark | 6% | USD106 |
Against cash, excess return is percentage points. Against the equity benchmark, it is percentage points. The latter is also called active return. Both statements are correct: they subtract different alternatives from the same portfolio return.
Original synthetic comparison. The top panel measures returns on equal starting capital. The lower panel shows why changing the denominator changes the reported measure. No costs, taxes or external cash flows are included.
“Five percentage points above cash” is different from saying that portfolio wealth is 5% larger than cash wealth. The portfolio finishes USD5 ahead, but cash finishes with USD103. The wealth-relative comparison is
The arithmetic excess return uses the gap per unit of initial capital; the wealth-relative measure uses the gap per unit of reference ending wealth. The latter interpretation requires positive reference wealth. For multiple periods, compound each investment’s returns before comparing cumulative outcomes; summing periodic arithmetic excess returns generally does not give the cumulative return difference.
Specify what went into the subtraction
A usable report names the reference instrument or index, the start and end dates, reporting currency, and return basis. “Above cash” should identify the cash proxy and the matching holding-period return. A quoted annual yield is not automatically the cash return over a shorter portfolio holding period.
Both sides should use compatible valuation times and distribution conventions. Comparing a portfolio including dividends with a price-only index confounds income with outperformance. State whether the comparison is nominal or inflation-adjusted, currency-hedged or unhedged, and before or after management fees, transaction costs, financing charges and taxes. A benchmark can be a useful opportunity-cost reference while excluding costs an investor would incur to track it; disclose that distinction.
For a concrete research convention, Kenneth French’s factor documentation defines the US market factor by subtracting its specified one-month Treasury-bill reference from the market return. This illustrates why the data provider’s definition matters: a column already labelled as an excess return should not have the same cash reference subtracted a second time.
Realised excess, expected premium and model alpha
The realised excess return is an observed difference. An expected risk premium is a forward-looking expectation of return above the risk-free reference, interpreted as compensation for bearing risk. A favourable realised period neither proves that the expected premium was high nor guarantees that it will persist.
Alpha adds a model of systematic exposure. In a single-market excess-return regression,
The left side is the portfolio’s excess return. The intercept is a model-dependent component after accounting for market exposure ; it is not the whole left side. One observed excess return cannot identify that intercept or demonstrate manager skill. The reference, risk model, estimation period and uncertainty all matter.
A historical Sharpe ratio uses the average of a series of cash-relative excess returns divided by their standard deviation. It asks about reward per unit of variability, which a single positive excess return cannot answer. The Risk-Adjusted Performance Measures lesson develops Sharpe and Jensen’s regression framework. A public preview is available; the full lesson requires membership.
Sources
- Edward E. Qian, Ronald H. Hua and Eric H. Sorensen, Quantitative Equity Portfolio Management: Modern Techniques and Applications, Chapman & Hall/CRC (2007), §2.2.2, p. 30, and §4.1, pp. 82–83 — expected returns above cash and performance measures using excess-return series. The example and figure here are original.
- Stephen Boyd et al., Multi-Period Trading via Convex Optimization, §3.2, pp. 21–22 — active return relative to a benchmark, the cash-relative special case, and information and Sharpe ratios.
- William F. Sharpe, The Sharpe Ratio, The Journal of Portfolio Management (Fall 1994), author-hosted reprint — expected versus historical differential returns, variability and limitations of historical inference.
- Kenneth R. French, Description of Fama/French Factors — the provider’s explicit US market excess-return convention.