4.4.2 The Merton Structural Model

Basic Prerequisites: 01 · From Zero and Black–Scholes–Merton.

---

1. Intuition & Practical Objective

The structural model takes the payoff insight of page 01 and turns it into a pricing machine. Two assumptions do all the work:

1. The firm’s assets follow a geometric Brownian motion, $dV=\mu Vdt+{\sigma }_{V}VdW$.
2. The firm has issued a single zero-coupon bond of face $D$ maturing at $T$; default happens only at $T$, and only if $V_T<D$.

Under those assumptions the equity claim $E=VN(d_1)-D{e}^{-rT}N(d_2)$ is not an approximation - it is the exact no-arbitrage price of the shareholders’ position. The practical objective is the inverse problem: real data gives us equity price $E_0$ and equity volatility ${\sigma }_E$, but the model runs on $(V_0,{\sigma }_V)$, which are unobservable. So we must solve the two-equation system

$$E_0=V_{0}N(d_1)-D{e}^{-rT}N(d_2),{\sigma }_E{E}_0=N(d_1){\sigma }_V{V}_0$$

for the two unknowns $(V_0,{\sigma }_V)$. Everything downstream - distance to default, PD, credit spread - is then deterministic algebra.

Why two equations? Equity fixes the scale of firm value (the call equation); equity volatility fixes the firm’s riskiness (the vol identity, which is Itô applied to $E(V,t)$: ${\sigma }_{E}E=\partial E/\partial V\cdot {\sigma }_{V}V=N(d_1){\sigma }_{V}V$).

---

2. Mathematical Ground Truth & Derivations

2.1 The equity PDE (Merton 1974, eqs. 10–11)

Let $f(V,t)$ be equity value and $F(V,t)$ the debt value, $V=f+F$. The firm’s asset dynamics give, for any claim on the firm’s assets under the risk-free measure,

$$\frac{1}{2}{\sigma }_V^2{V}^2{F}_{VV}+rV{F}_V-rF-F_t=0.$$

Substituting $f=V-F$ and using $F_{VV}=-f_{VV}$, $F_V=1-f_V$ (Merton eq. 15), the equity value satisfies the same BSM PDE,

$$\frac{1}{2}{\sigma }_V^2{V}^2{f}_{VV}+rV{f}_V-rf-f_t=0,$$

with terminal condition $f(V,0)=\max (0,V-B)$ and boundary conditions $f(0,t)=0$, $f(V,t)\le V$. This is precisely the BSM PDE for a European call on $V$ struck at $B$ - so the solution is (Merton eq. 12)

$$\boxed{E=f(V,t)=VN(d_1)-B{e}^{-rt}N(d_2)},d_1=\frac{\ln (V/B)+(r+\frac{1}{2}{\sigma }_V^2)t}{{\sigma }_V\sqrt{t}},d_2=d_1-{\sigma }_V\sqrt{t}.$$

2.2 The debt value and the risk structure (Merton eqs. 13–14)

From $F=V-E$,

$$F(V,t)=B{e}^{-rt}\left\{\Phi \right[h_2(d,{\sigma }_V^{2}t)]+\frac{1}{d}\Phi [h_1(d,{\sigma }_V^{2}t)\left]\right\},$$

where the quasi debt-to-firm ratio is $d=B{e}^{-rt}/V$ and

$$h_1(d,{\sigma }_V^{2}t)=-\frac{\frac{1}{2}{\sigma }_V^{2}t-\ln d}{{\sigma }_V\sqrt{t}},h_2(d,{\sigma }_V^{2}t)=-\frac{\frac{1}{2}{\sigma }_V^{2}t+\ln d}{{\sigma }_V\sqrt{t}}.$$

Because bond talk is in yields, define the risky yield by $e^{-R(t)t}=F(V,t)/B$, giving Merton’s risk structure of interest rates (eq. 14):

$$R(t)-r=-\frac{1}{t}\ln \left\{\Phi \right[h_2(d,{\sigma }_V^{2}t)]+\frac{1}{d}\Phi [h_1(d,{\sigma }_V^{2}t)\left]\right\}.$$

The striking content: for a given maturity, the credit spread depends on only two variables - the business risk ${\sigma }_V^2$ and the quasi leverage $d=B{e}^{-rt}/V$.

2.3 Comparative statics (Merton eq. 15)

With $F[V,t,B,{\sigma }_V^2,r]$ shown in full:

$$F_V=1-f_V>0,F_B=-f_B>0,F_t=-f_t<0,F_{{\sigma }_V^2}=-f_{{\sigma }_V^2}<0,F_r=-tD{e}^{-rt}N(d_2)<0.$$

Translation: debt value rises with firm value and promised payment; falls with time to maturity, firm volatility, and the riskless rate. Since equity is a call and the call is homogeneous of degree one and convex in $(V,B)$, default-free debt-plus-equity is concave in $V$ - the origin of the equity skew.

2.4 The two-equation inversion (Hull eqs. 24.3–24.4)

The observable system is Hull’s §24.6:

$$\boxed{E_0=V_{0}N(d_1)-D{e}^{-rT}N(d_2)},\boxed{{\sigma }_E{E}_0=N(d_1){\sigma }_V{V}_0},$$

with $\mathbb{Q}(\text{default})=N(-d_2)$. This is solved numerically (fixed point or Newton) - §3 below.

---

3. Computational Implementation - forward map, then inverse map

The acid test of any structural-model implementation: assume true firm parameters, generate the equity observables, then feed only the equity observables back in and check that the solver recovers the firm parameters. Stdlib only.

The solver recovers the firm parameters to machine precision - the forward/inverse maps are consistent. Note the last line is already a failure-mode preview: the model spread is $2.7$ bp for a firm whose equity vol is $48\%$ and whose book leverage ($D/V=0.6$) is high.

---

4. Failure Modes & First-Principles Breakdowns

1. The inverse problem is ill-conditioned. The two-equation system must be solved numerically, and near the interesting region (levered, high-vol firms) the Jacobian is nearly singular - small errors in ${\sigma }_E$ produce large errors in ${\sigma }_V$ and hence in PD. See 03 and 05.
2. Single zero-coupon bond, default only at $T$. Real capital structures are multi-layered and default can happen at any time. Merton’s own §VI extends to coupon/callable bonds, but the vanilla formula’s “$T$-only” default is the source of the credit-spread puzzle.
3. Constant ${\sigma }_V$ and a lognormal firm. The same constant-volatility delusion that afflicts BSM afflicts the firm: no jumps means no jump-to-default, so a firm can never default “suddenly”.
4. $\partial E/\partial V=N(d_1)<1$ breaks the naive hedge. Equity is a levered claim: equity vol exceeds asset vol, and equity beta with the market exceeds 1 - a direct consequence of the convexity in §2.3 that practitioners must respect.

---

5. References

• Merton (1974), On the Pricing of Corporate Debt
• Hull, Options, Futures, and Other Derivatives
• Bluhm, Overbeck & Wagner, Introduction to Credit Risk Modeling

---

6. Connected Graph Bridges

• Back: 01 · From Zero
• Forward: 03 · Distance-to-Default & PD · Index Hub
• Theory: Black–Scholes–Merton · 06 · Portfolio Credit & Vasicek