Companies that consider linking equity awards to performance should prepare to dig in for deeper computations of the compensation's fair value.
For ordinary options, fair-value calculations are straightforward, with values plugged in from either the Black-Scholes options-pricing model or a lattice model to produce the result. But when vesting is contingent on a relative market condition — how much the growth of company total shareholder return exceeds TSR for a group of peers, for example — the computation of fair value must include the probability that options won't vest. And for that, the standard models usually won't do.
The reason is that both the Black-Scholes and lattice models reflect a small number of variables, including current stock price, option strike price, and share volatility. But when vesting contains a time element, and such factors as a peer group's average stock-price volatility compared with the company's volatility, a more flexible approach is needed.
Typically, that calls for a Monte Carlo simulation, allowing analysts to build models with many variables. A computer creates a forecast by generating random values for the variables and then running the model thousands of times to create a probability distribution. The most difficult part is the modeling. "It's time-consuming to build the model from scratch," says Stacy Powell of actuarial consulting firm CCA Strategies, "but once you have it, it's not much more difficult than using the lattice method."
Companies have more leeway basing compensation awards on internal company goals. Instead of prescribing a certain method, FAS 123R says that companies must show they are using their best estimate of the likelihood of a payout. For an earnings-per-share-linked goal, that might mean using the firm's own forecasting methodology. For nonfinancial goals, it might be little more than an educated guess. "What's important is that you disclose any assumptions that went into the probabilities you're using," says Powell. — D.D.
Reader Comments» Post a comment