This dissertation consists of three empirical essays regarding the estimate of future market volatility implied by an option price. According to the theory of market efficiency, this implied volatility should represent the market participants' consensus expectation of the volatility over the remaining life of the option. But this assertion has been hotly debated. The first essay presents a new explanation for the previous finding that implied volatility is biased and inefficient. I find that implied volatility varies around the market's true volatility expectation due to measurement error and that this measurement error biases tests towards rejecting the informational efficiency of the implied volatility. When I control for the measurement error utilizing instrumental variables estimation, the results in most cases no longer reject the hypothesis that implied volatility is unbiased and informationally efficient.

The second essay relates to the implied volatility "smile". While previous explanations for the smile have focused on possible errors in the pricing formula, I argue that the smile may be caused by investors' preferences for certain strike prices for hedging purposes. I find that market inefficiency is partly responsible for the smile since abnormal returns can be made over time based on the implied volatility differences. The third essay compares the relative forecasting efficiency of implied volatilities across different strike prices. I find, contrary to the general belief and practice, that the implied volatility calculated from an at-the-money option is less informative than those calculated from options with relatively higher strike prices and that an average measure may not be effective.