Equity premium puzzle¶

Confront one number against one model — the ~6-point historical equity premium against what a consumption-CAPM with plausible risk aversion can rationalize — and read the order-of-magnitude miss as indicting a load-bearing assumption in an enumerable stack.

Version: v4 · 2026-09-19 · History
Domain-specific #: 348
Domain group: Social Sciences
Origin domain: Economics & Finance
Subdomain: Asset Pricing → Economics & Finance

Core Idea¶

The equity premium puzzle, formalized by Mehra and Prescott (1985), is that the historical excess return of US equities over bonds — roughly 6 points per year — is orders of magnitude larger than a standard consumption-based asset-pricing model can rationalize at plausible risk aversion. The model prices the premium as compensation for equity's covariance with consumption growth; because that growth is smooth and weakly correlated, matching the premium demands implausible risk aversion of 30-40.

How would you explain it like I'm…

The Too-Big Reward Mystery

Grown-ups can save money in a safe place or in a bumpier place called stocks. The bumpy place paid way more extra money over many years. Experts' best math says people shouldn't need that much extra, because their everyday spending didn't go up and down very much. Nobody is sure why the gap is so big, so it's called a puzzle.

The Too-Big Stock Bonus Mystery

People can invest in stocks, which are risky, or in short-term government bonds, which are safe. Historically in the US, stocks earned a lot more than bonds each year; that extra is called the equity premium. Economists have a standard model that says the extra should be a reward for risk, measured by how much stocks go up and down along with what people spend. But people's spending changed fairly smoothly, so the model predicts a much smaller extra than what really happened. To make the model fit, people would have to be wildly afraid of risk. Economists have proposed many explanations, but none has won everyone over.

Unexplained Stock-Bond Return Gap

The equity premium puzzle was formalized by Mehra and Prescott in 1985. From 1889 to 1978, US stocks beat short-term government bonds by about 6 percentage points a year on average. A standard model, in which one representative person picks investments to smooth their consumption, says stocks deserve extra return only to the extent that their returns move with consumption. US consumption has been smooth and only weakly tied to stock returns. So the model can match the 6% gap only if people are wildly risk-averse, far beyond the usual assumed range. The puzzle is about the size of the gap, not about stocks earning high returns. Proposed explanations include different preferences, rare disasters, limits on who invests in stocks, and survivorship bias, but none has won general agreement.

Mehra and Prescott (1985) showed that the historical excess return of US equities over short-term government bonds—roughly 6 percentage points per year over 1889–1978—is an order of magnitude larger than a standard consumption-based asset-pricing model can rationalize. That model is the representative-agent consumption CAPM with CRRA preferences, in which the premium compensates for the covariance of equity returns with aggregate consumption growth. Because postwar US consumption growth has low volatility and weak correlation with stock returns, matching the premium requires relative risk aversion of roughly 18 for the premium alone, up to about 55 when also matching the low risk-free rate, versus a commonly assumed range of 1–10. The puzzle therefore concerns the size of the gap relative to consumption risk: either agents are extraordinarily risk-averse, consumption data misstate the risk they face, or expected utility is wrong. Responses include preference-side models (habit formation, Epstein–Zin recursive utility separating risk aversion from intertemporal substitution, loss and ambiguity aversion), rare-disaster and tail-risk arguments, limited-participation and friction stories in which the marginal equity investor has more volatile, equity-correlated consumption, and survivorship bias in using the US series. None has achieved consensus, and it remains the central unsolved calibration anomaly in asset pricing.

Scope of Application¶

The equity premium puzzle is a single calibration anomaly scoped to asset pricing, not a recurring mechanism; its reach is depth on one number within a small cluster of consumption-CAPM anomalies.

Consumption-based asset pricing — the home turf: the central unsolved calibration target.
The risk-free-rate puzzle — its mirror image, the same anomaly from the other leg.
Preference-theory research — habit formation, Epstein-Zin, loss aversion, ambiguity aversion.
Rare-disaster macro-finance — the branch arguing the postwar sample omits tail risk.
Limited-participation asset pricing — denying the representative consumer is the marginal investor.
Long-run return measurement — the survivorship-bias branch on the US series.

Clarity¶

The puzzle relocates the question, converting the vague "equities are risky, so they pay more" into a calibration: one number the model must reproduce from constrained inputs. Naming it makes the order-of-magnitude gap a fact to be explained, and forces every proposed fix to declare which load-bearing assumption it abandons, so rival resolutions can be compared on common ground.

Manages Complexity¶

Ninety years of returns on hundreds of securities collapse to one confrontation between one number and one model. That compression imposes a fixed branch structure: because the failing model rests on a short, enumerable assumption stack, every resolution must abandon at least one named element — preferences, the consumption tail, the marginal investor, or the data series.

Abstract Reasoning¶

The puzzle licenses diagnostic backward inference from the size of the miss, assumption-localization that reads each fix as a wager on one layer of an explicit stack, interventionist derivation of differential empirical predictions from each branch, and boundary-drawing that keeps the persistent premium from being misclassified as arbitrage.

Knowledge Transfer¶

Within asset pricing the puzzle transfers as a live research object — the full apparatus carries — but it is one anomaly, not a recurring mechanism; its reach is depth, not breadth. Beyond it, moving the name strips the model and data (analogy). The genuinely portable content is the parent: a calibrated model diverging from a measured quantity beyond tuning, indicting an assumption, input, or frame. Carry that parent, not the consumption-covariance furniture.

Relationships to Other Abstractions¶

Current abstraction Equity premium puzzle Domain-specific

Parents (1) — outgoing typed relations

Equity premium puzzle is a decomposition of Calibration Anomaly Prime

The Equity Premium Puzzle is the asset-pricing form of a calibration anomaly in which an independently constrained model misses an independently measured target beyond plausible retuning.

Children (1) — abstractions linked here by typed relations

Risk-Free Rate Puzzle Domain-specific presupposes Equity premium puzzle

The Risk-Free Rate Puzzle arises when raising CRRA risk aversion to repair the Equity Premium Puzzle drives the same model's risk-free-rate prediction implausibly high.

Hierarchy path (1) — routes to 1 parentless root

Equity premium puzzle → is a decomposition of Calibration Anomaly

Neighborhood in Abstraction Space¶

Equity premium puzzle sits in a moderately populated region (52^nd percentile for distinctiveness): it has near-neighbors but no dense thicket of look-alikes.

Family — Unclustered & Miscellaneous (2769 abstractions)

Nearest neighbors

Computed from structural-signature embeddings · 2026-10-25