Bjerksund-Stensland Model

The Bjerksund-Stensland model is an analytical closed-form approximation for pricing American call and put options on dividend-paying equities. Instead of solving a full partial differential equation or building a binomial lattice, it uses a clever boundary approximation that treats early exercise as occurring whenever the stock price hits a certain optimal level. The result is fast, accurate to within 1–2 basis points, and widely adopted by practitioners who need option prices in real time.

The American option problem: why closed-form is hard

A European option can only be exercised at maturity, so the Black-Scholes-Merton formula applies: an integral under the risk-neutral measure, solvable in closed form. An American option can be exercised at any time before expiry, opening the door to early exercise. This optimal stopping problem has no known closed-form solution—the value at any point depends on whether immediate exercise is better than waiting, a recursive decision that typically demands numerical methods.

The standard approach is a binomial tree: at each node, you decide whether to exercise now or hold and optimize forward. For fine grids (thousands of steps), this becomes slow—problematic for a trader who needs prices for hundreds of option strikes in milliseconds. The Bjerksund-Stensland approximation offers a compromise: a formula that captures the essential logic of early exercise without building the tree.

The key insight: approximating the exercise boundary

The crux is the optimal early-exercise boundary: a stock price S(t) as a function of time and parameters, such that you should exercise the American option if and only if S hits this boundary. Bjerksund and Stensland observe that for perpetual American options (infinitely long maturity), the boundary is constant: above that price, always exercise; below it, always hold. For finite maturity, the boundary is not constant but can be approximated.

Their trick: use the perpetual-option boundary as a first guess, then apply a correction for time-to-maturity and dividend yield. This yields a closed-form approximation for the boundary, which in turn gives a closed-form value.

The formula in sketch (calls on dividend-paying stock)

For an American call on a stock paying continuous dividend yield q, with strike-price K, stock price S, volatility σ, interest-rate r, and time to maturity T:

1. Compute a “trigger” price S (boundary) using the perpetual-option formula scaled by a time-dependent factor.
2. If the current stock price exceeds the trigger and early exercise is optimal, use the immediate exercise-price plus time value.
3. Otherwise, use a Black-Scholes-Merton-like integral adjusted for the boundary.

The full formula is a two-piece structure: one approximation for the value itself, and one for whether early exercise is in the money. The result is a function you can evaluate in microseconds—a handful of exponentials and square roots.

Accuracy and when it fails

The Bjerksund-Stensland approximation is typically accurate to 1–2 basis points for option prices in the range of 0.5 to 1.5 times the strike-price, and less accurate (errors of 5–10 basis points) when deep in or out of the money. The error stems from the approximation of the true boundary: the real boundary is more nuanced than the simplified formula captures, especially near maturity (when the boundary is close to the strike-price).

For portfolios with millions of option positions and tight margin constraints, traders sometimes prefer the Barone-Adesi Whaley model, which uses a quadratic approximation and can be even faster. For academic research or pricing where accuracy is paramount, a full binomial or finite-difference lattice is more reliable.

Dividend yield and adjustments

The original Bjerksund-Stensland formula assumes a constant continuous dividend yield q. If dividends are discrete (e.g., a known dollar amount paid on a specific date), the formula becomes less accurate, though practitioners use adjustments or approximate the discrete stream as a continuous yield.

For stock options in markets with ex-dividend dates (essentially all real-world equity option markets), traders price American options using Bjerksund-Stensland as a baseline and then adjust for discrete dividends via a shift in the stock price or a manual check on the optimal exercise-price around dividend dates.

Comparison with other methods

• Binomial trees: Exact in the limit, but slow (thousands of nodes). Bjerksund-Stensland is much faster.
• Finite-difference PDE solvers: Also exact (given fine grids), but require specialized software and calibration. Bjerksund-Stensland is closed-form.
• Barone-Adesi Whaley model: A competing approximation using quadratic terms instead of boundary approximation. Often slightly more accurate for deep in-the-money options.
• Monte Carlo: Flexible for complex payoffs but slow for American options (requires Longstaff-Schwartz regression or similar).

See also