Equity Premium Puzzle

The equity premium puzzle is a gap between theory and reality: stocks have returned roughly 7–8% per year above risk-free Treasury rates for over a century, yet standard models predict the spread should be only 1–2%. No satisfying explanation for this discrepancy exists, making it one of finance’s most stubborn puzzles.

The measurement

Start with the plain facts. From 1926 to the present, US equities have returned approximately 10% annually (including dividends). Risk-free Treasury bills have returned about 3%. The difference—the equity premium—sits at roughly 7%. Over a century, this spread compounds into enormous wealth differences between stock and bond portfolios.

Now consider risk. Stocks are volatile; their returns swing wildly. Treasury bills are stable. Standard economic theory says investors should demand compensation for bearing risk, and the premium is exactly that compensation. The puzzle is not that the premium exists—it should. The puzzle is its magnitude.

Mehra and Prescott’s 1985 calculation is the canonical statement. Using standard economic models of consumption and risk aversion, they asked: what equity premium does theory predict? Their answer: roughly 0.35% per year, possibly stretching to 1% under generous assumptions. The observed premium of 7% is 7–20 times larger. No tweak to standard models closes the gap.

Why theory fails

The workhorse model assumes investors care about consumption. They want enough purchasing power to buy food, shelter, and goods. Risk matters to the extent it threatens consumption. A stock that crashes when the economy booms is less risky than one that crashes during recessions—the former lets you substitute leisure for consumption, whilst the latter force it on you unwillingly.

Given this framework, how risky are stocks? The answer is surprisingly small. Stock returns and consumption growth are only weakly correlated. A bad year in the stock market does not reliably predict a bad year for economic growth or consumption. Therefore, stocks should not carry a large premium.

Yet they do. Either the model is missing something fundamental about risk, or investors are not behaving as the model assumes.

Proposed resolutions—and their limits

Consumption volatility. Some researchers argue that consumption is more volatile than measured GDP suggests. If true, stocks would be riskier than it appears, justifying a higher premium. The problem: efforts to measure true consumption volatility find no improvement over the baseline. The puzzle remains.

Survival risk. Perhaps investors fear economic collapse—a scenario where stocks plummet and stay depressed for decades. Standard models assume risk is normally distributed; extreme tail events are ignored. If tail risk is real and large, the premium compensates for it. But tail events are so rare that accounting for them statistically should matter little. And yet stocks underperform these models’ predictions even during normal years.

Habit formation. Investors might care not just about absolute consumption but about consumption relative to a habit level. A drop below habit creates outsized pain. If so, even modest consumption volatility triggers large risk premiums. Some versions of this model can generate an equity premium in the right ballpark, but only by assuming implausibly high aversion to falling below habit, and the evidence that people behave this way is mixed.

International evidence. The puzzle does not vanish in other countries. Japanese, UK, and European stock markets also show equity premiums larger than theory predicts—sometimes larger than the US premium, despite lower expected growth. This suggests the puzzle is not a US artifact or data-measurement problem.

Behavioral explanations

If models with rational investors fail, perhaps investors are not rational. Several psychological factors could inflate the equity premium.

Loss aversion. Investors feel losses roughly twice as acutely as equivalent gains. Stocks offer high upside potential but severe downside; bonds offer steady, modest returns. Loss-averse investors may shun stocks not because they are rationally too risky but because the prospect of loss feels unbearable. To attract them, stocks must offer a giant premium. This explains the magnitude but raises a new question: why would a competitive market allow such permanent mispricing?

Overconfidence and overtrading. If retail investors are overconfident in their ability to pick winners, they may overshoot into stocks, driving prices up and expected returns down. Simultaneously, institutions shunning stocks (perhaps due to career risk) may exacerbate the shortage. The result is a non-clearing market where stocks trade at prices implying lower equilibrium returns than the models predict.

Myopia and mental accounting. Investors might focus on short-term volatility rather than long-term returns. A stock that rises 50% then falls 40% looks terrifying month-by-month but stellar over a decade. If investors weight near-term pain heavily, they demand an outsized premium. Alternatively, investors may compartmentalise—treating stock and bond portfolios separately—and thus avoid the insight that bonds are redundant insurance within a long-term horizon.

Rare disasters. Stocks may underperform in the rare event of economic ruin (war, pandemic, collapse). If investors fear this, they will demand a large premium not proportional to baseline risk. This is theoretically coherent but difficult to test and requires assuming investors weight low-probability catastrophes far more than historical frequency suggests they should.

Unresolved questions

After 40 years, no consensus answer exists. Most likely, the true explanation involves both omitted risks (consumption volatility, rare disasters, or risks not yet identified) and behavioral factors (loss aversion, myopia, market frictions). The fact that the puzzle persists suggests that markets do not perfectly price equities, at least at the very long end.

For practitioners, the puzzle has practical consequences. If the observed equity premium is not a “true” risk premium but rather a mispricing or compensation for behavioral factors, it may not be sustainable. A shift in risk tolerance or a new generation of rational investors could rapidly compress the premium. Conversely, if the high premium truly compensates for deep risks (consumption shocks, tail events), it is fair, and equity allocations should remain generous.

The puzzle also highlights the limits of theory. Finance models are elegant but sometimes divorced from reality. The fact that they cannot explain the largest and most important number in asset pricing—the return difference between stocks and bonds—is humbling.

See also