The Merton 1974 model¶
Merton (1974) treats firm equity as a European call option on the firm’s underlying assets.
Setup¶
A firm has total asset value \(A_t\) that follows geometric Brownian motion under the risk-neutral measure:
At maturity \(T\) the firm owes a single liability \(D\).
If \(A_T \geq D\), equity holders receive \(A_T - D\) and bondholders are paid in full.
If \(A_T < D\), equity holders walk away (limited liability); the firm defaults; bondholders receive \(A_T\).
Equity therefore has the payoff \(E_T = \max(A_T - D, 0)\) — a European call.
Equity value via Black-Scholes-Merton¶
with
Distance to default and PD¶
Define the distance to default as \(d_2\):
The risk-neutral probability of default is
Converting to the physical measure uses the asset-class Sharpe ratio \(\lambda = (\mu - r)/\sigma\):
Implied credit spread¶
For a horizon \(T\) and loss-given-default \(\mathrm{LGD}\), the implied continuous-compounding spread is
Inferring \(A_0\) and \(\sigma_A\)¶
Both are unobservable. The classic Jones-Mason-Rosenfeld system uses the observed equity value \(E_0\) and equity volatility \(\sigma_E\) together with Itô’s lemma to solve
for the two unknowns \(A_0\) and \(\sigma_A\). Vassalou-Xing extend the procedure to use the full equity time series under maximum likelihood; the snapshot case collapses back to the JMR two-equation system.
Limitations¶
A single debt maturity is assumed (Geske handles compound debt structures).
The default barrier is reached only at \(T\) (Black-Cox allows touch-default at any \(t \leq T\)).
The asset process is pure Brownian (Zhou’s jump-diffusion adds jumps).
Interest rates are constant (Longstaff-Schwartz adds stochastic rates).
All values are assumed to be lognormal (CreditGrades, Leland-Toft relax this).
All of these refinements ship under merton.extensions.