This paper studies the "overpriced puts puzzle" - the finding that historical prices of the S&P 500 put options have been too high and incompatible with the canonical asset-pricing models, such as CAPM and Rubinstein (1976) model. Simple trading strategies that involve selling at-the-money and out-of-the-money puts would have earned extraordinary profits. To investigate whether put returns could be rationalized by another, possibly nonstandard equilibrium model, we implement a new methodology. The methodology is "model-free" in the sense that it requires no parametric assumptions on investors' preferences. Furthermore, the methodology can be applied even when the sample is affected by certain selection biases (such as the Peso problem) and when investors' beliefs are incorrect.
We find that no model within a fairly broad class of models can possibly explain the put anomaly.
So, how likely is the "hosing"? Does this merely reflect the risk of large losses? By his estimates, there would have to be a meltdown like the one in October 1987 1.3 times a year for the option writer to lose money.
So, why are put options so apparently overvalued? There are at least two possible explanations (other than something really funky/wrong with the data): one is that investors systematically overestimate the chance or severity of large market declines. The other is that option buyers have a utility function that is extremely risk averse. In either case, there's apparently an excess demand for insurance that option writers can benefit from (if they're willing to bear the risk).
HT: CXO Advisory group