The reliability of employee stock option fair value disclosures under SFAS 123

This dissertation empirically examines whether currently available option-pricing models can accurately estimate the true cost of an employee stock option (ESO). This research question is motivated by the release of SFAS 123, which requires firms to disclose ESO fair values in the footnotes of their financial statements. ESO fair values must be estimated using an option-pricing model, though little is known about the validity of these models in pricing ESOs.; Because market prices do not exist for ESOs, the traditional avenues of validity testing of option-pricing models are unavailable. The methodology employed in this study draws on techniques from the economic forecasting literature. ESO model values are viewed as forecasts of the option's terminal value at exercise. Predicted values are then contrasted with actual exercise values, which are discounted for the time value of money and for risk. Model validity is assessed by applying Theil's tests of forecast rationality in a variance components framework.; The results suggest that current option-pricing models significantly overestimate ESO values, even when ESO fair values are estimated using expected maturities rather than stated maturities. Models that incorporate stochastic interest rates or stochastic stock return volatility are no more accurate than a simple Black-Scholes model. The amount of overstatement is roughly 15-20% and is more pronounced for firms with high stock return volatilities, high systematic risk, and low dilution from ESO issues. These biases in ESO fair value estimates should be taken into account when interpreting the disclosures required under SFAS 123.