Key Takeaway
Over the past several decades, S&P 500 implied volatility has exceeded subsequent realized volatility roughly 90% of the time. This persistent gap -- the variance risk premium -- means that selling options or variance swaps has generated positive expected returns, compensating investors for bearing the risk of rare but severe market crashes. The premium is one of the most robust and well-documented anomalies in derivatives markets, but harvesting it demands respect for the tail risks that justify its existence.
The Gap Between Fear and Reality
Between 1990 and 2023, the VIX -- the market's forward-looking estimate of 30-day S&P 500 volatility derived from option prices -- averaged approximately 19.5 percentage points. Over the same period, realized volatility on the S&P 500 averaged roughly 15.4 percentage points. That difference of about 4 volatility points, sustained over more than three decades, represents an enormous cumulative transfer of wealth from option buyers to option sellers.
Why does this gap persist? The answer lies in a fundamental asymmetry: most market participants need portfolio insurance far more urgently than they need to sell it. Pension funds, mutual funds, and structured product issuers are natural buyers of downside protection. They are willing to pay above actuarial fair value because the consequences of an unhedged crash -- forced liquidations, margin calls, career risk for portfolio managers -- are catastrophic in ways that cannot be captured by expected return calculations alone.
Carr and Wu (2009) formalized this observation. Using synthetic variance swap rates across a broad panel of individual stocks and equity indices, they demonstrated that the variance risk premium is consistently negative -- meaning the market price of variance systematically exceeds its expected future value. The premium was not confined to equities: it appeared across asset classes and persisted after controlling for standard risk factors.
Defining the Variance Risk Premium
The variance risk premium (VRP) is defined as the difference between risk-neutral expected variance (implied from option prices) and the physical expected variance (what volatility actually materializes):
VRP = Implied Variance - Realized Variance
When VRP is positive, option sellers collect more premium than the actual volatility they experience would justify. The premium is most commonly measured using the VIX index (or its squared value for variance) minus subsequent realized variance over matching horizons.
The following table summarizes the typical magnitude of the VRP across different market conditions, based on data from 1990 through 2023:
| Market Regime | Avg. Implied Vol (VIX) | Avg. Realized Vol | VRP (vol points) | VRP as % of Implied |
|---|---|---|---|---|
| Low volatility (VIX < 15) | 13.2 | 9.8 | 3.4 | 26% |
| Normal (VIX 15-25) | 19.1 | 15.3 | 3.8 | 20% |
| Elevated (VIX 25-35) | 28.7 | 23.9 | 4.8 | 17% |
| Crisis (VIX > 35) | 48.3 | 41.6 | 6.7 | 14% |
| Full sample average | 19.5 | 15.4 | 4.1 | 21% |
Two patterns stand out. First, the premium exists in every regime -- even during crises, implied volatility overshoots realized volatility on average. Second, the absolute magnitude of the premium increases with the level of fear, though it shrinks as a percentage of implied volatility. During crises, the market overestimates future volatility by more in absolute terms but by less in proportional terms, because realized volatility itself rises sharply.
Why the Premium Exists: Insurance Economics
Carr and Wu (2009) documented that the variance risk premium cannot be explained by standard asset pricing models. It is not compensation for equity market risk (beta), nor is it captured by the Fama-French factors. Instead, it behaves like an insurance premium embedded in option markets.
The analogy is precise. Homeowners pay more for fire insurance than the actuarial expected loss from fires because the consequences of an uninsured fire are disproportionately severe. Similarly, portfolio managers pay more for downside protection than the expected payout because an unhedged crash can trigger cascading consequences -- margin calls, fund redemptions, regulatory breaches, and career destruction -- that far exceed the direct financial loss.
Ilmanen (2012) placed the VRP within a broader framework. Across multiple asset classes and instrument types, he documented that selling insurance-like payoffs -- strategies with negative skewness and high kurtosis that collect steady premiums punctuated by occasional large losses -- earns positive risk-adjusted returns. Conversely, buying lottery-like payoffs -- strategies that offer small chances of large gains funded by steady losses -- earns negative risk-adjusted returns. The VRP is the single clearest expression of this pattern in liquid financial markets.
Harvesting the Premium in Practice
The pure academic expression of the VRP involves variance swaps -- contracts that pay the difference between realized and implied variance. In practice, most investors access the premium through simpler instruments:
Selling index put options. The most common approach. Selling out-of-the-money puts on the S&P 500 -- typically 5% to 10% below current levels with 30 to 45 days to expiration -- collects the volatility risk premium while maintaining a defined risk profile. The CBOE PutWrite Index (PUT), which tracks a strategy of selling cash-secured at-the-money puts on the S&P 500, has delivered equity-like returns with lower volatility over its backtest history.
Short straddles and strangles. Selling both puts and calls captures the VRP on both sides but exposes the seller to losses from large moves in either direction.
VIX futures roll. The VIX futures term structure is typically in contango, meaning longer-dated futures trade above the spot VIX. Shorting VIX futures and rolling positions captures the convergence of futures prices toward spot as expiration approaches -- effectively harvesting the same premium from the futures side.
The Catch: Tail Risk and Left-Tail Exposure
The variance risk premium exists precisely because harvesting it is painful during crises. Selling volatility is the financial equivalent of selling earthquake insurance: years of steady premium income punctuated by occasional devastating claims.
During the 2008 financial crisis, the VIX spiked from 20 to 80 in a matter of weeks. A systematic put-selling strategy would have suffered drawdowns of 30% to 40% in a single quarter. The March 2020 COVID crash produced a VIX spike from 14 to 82 in less than a month. In February 2018, the implosion of short-volatility products linked to the VIX -- most notably the XIV exchange-traded note -- demonstrated that leveraged volatility selling can produce total losses in a single session.
These episodes are not anomalies; they are the reason the premium exists. As Ilmanen (2012) emphasized, the positive expected returns from selling insurance are compensation for bearing precisely these catastrophic scenarios. An investor who cannot survive the drawdown cannot harvest the premium.
Sizing and Risk Management
The practical challenge is position sizing. Selling too much volatility relative to portfolio capital transforms a positive expected-value strategy into a ruin-risk strategy. Practitioners generally follow several guidelines:
Cap notional exposure. Limit the notional value of options sold to a fraction of total portfolio value -- typically 25% to 50% for an unleveraged approach. This ensures that even a 2008-style crash produces a painful but survivable drawdown.
Use spreads, not naked positions. Selling put spreads (sell a put, buy a further OTM put) caps maximum loss per contract. The purchased put sacrifices some premium income but eliminates the risk of catastrophic loss from a market gap far below the strike.
Diversify across tenors and strikes. Rolling positions across multiple expiration dates and strike levels reduces concentration risk and smooths the P&L profile.
Maintain a volatility-aware overlay. Reduce exposure when the VIX is unusually low (below 12-13), as the premium earned is small relative to the tail risk accepted, and the probability of a volatility spike is elevated from compressed levels.
Time-Varying Nature of the Premium
The VRP is not constant. Carr and Wu (2009) showed that the premium varies systematically with the level of market fear and macroeconomic uncertainty. It widens during recessions and risk-off episodes and narrows during extended low-volatility periods.
This time variation creates an opportunity for dynamic strategies. When the VIX is elevated -- above 25 or 30 -- the spread between implied and realized volatility tends to be at its widest, and selling volatility offers the most attractive risk-reward. When the VIX is compressed below historical norms, the premium is thin and the asymmetry between potential gains and losses is unfavorable.
However, timing the VRP is difficult. The same periods of elevated fear that offer the widest premiums are also the periods where tail events are most likely to occur. Systematic, rules-based harvesting with disciplined position sizing tends to outperform discretionary timing approaches over long horizons.
Actionable Takeaway
The variance risk premium is one of the most persistent and well-documented sources of return in financial markets. Selling volatility -- whether through put options, variance swaps, or short VIX futures -- has generated positive expected returns over decades because it provides insurance to a market that structurally demands crash protection. But the premium is compensation for real risk. Position sizing must account for the possibility of 30% to 50% drawdowns during extreme events. The most robust approach combines systematic premium collection with strict exposure limits, spread structures to cap tail losses, and the discipline to maintain the strategy through the periodic crises that make most participants abandon it.
References
-
Carr, P., & Wu, L. (2009). "Variance Risk Premiums." The Review of Financial Studies, 22(3), 1311-1341. https://doi.org/10.1093/rfs/hhn038
-
Ilmanen, A. (2012). "Do Financial Markets Reward Buying or Selling Insurance and Lottery Tickets?" Financial Analysts Journal, 68(5), 26-36. https://doi.org/10.2469/faj.v68.n5.7