The Envelope Theorem

Oscar Volij Iowa State University

The Envelope Theorem – p. 1/1

The Envelope Theorem – p. 2/1 .The unconstrained case Consider a simple maximization problem max f (x. θ) x (1) where x is a choice variable and θ is a parameter that we do not control. We assume that f is a concave function of x so that the first order conditions are not only necessary but sufficient for maximization.

θ) = 0 → x∗ (θ) ∂x (2) Let’s plug this optimal value into the objective function and define the optimal value of the objective function as a function of θ. V (θ) ≡ f (x∗ (θ). θ).First Order Conditions An interior solution of (1) satisfies the following first order conditions: ∂f (x. The Envelope Theorem – p. 3/1 .

Formally. we are interested in dV (θ). 4/1 . The Envelope Theorem – p. dθ Note that V will change both because θ affects f and because it also affect the optimal choice of x.The question We are interested in what happens to the optimal value of f when the parameter θ changes.

θ) dθ ∂x ∂θ ∂θ The Envelope Theorem – p.The Answer To answer this question we take derivatives: ∂x∗ dV ∂f ∗ ∂f ∗ (θ) = (x (θ). 5/1 . θ) (θ) + (x (θ).

5/1 . θ) (θ) + (x (θ). θ) dθ ∂x ∂θ ∂θ 0 The Envelope Theorem – p.The Answer To answer this question we take derivatives: ∂x∗ dV ∂f ∗ ∂f ∗ (θ) = (x (θ).

θ) dθ ∂x ∂θ ∂θ 0 ∂f ∗ (x (θ). 5/1 . θ) (θ) + (x (θ).The Answer To answer this question we take derivatives: ∂x∗ dV ∂f ∗ ∂f ∗ (θ) = (x (θ). θ) = ∂θ The Envelope Theorem – p.

The Answer To answer this question we take derivatives: ∂x∗ dV ∂f ∗ ∂f ∗ (θ) = (x (θ). θ) = ∂θ where the second equality follows from (2). θ) (θ) + (x (θ). θ) dθ ∂x ∂θ ∂θ 0 ∂f ∗ (x (θ). The Envelope Theorem – p. 5/1 .

The Envelope Theorem This proves the simple version of the envelope theorem: the total rate of change in the optimal value of the objective function due to a small change in the parameter θ is simply the rate of change in the objective function f evaluated at the optimal value of x. dV ∂f ∗ (θ) = (x (θ). 6/1 . θ) dθ ∂θ The Envelope Theorem – p.

The constrained case Consider a simple maximization problem max f (x. θ) ≤ 0 where x is a choice variable and θ is a parameter that we do not control. θ) x (3) s. 7/1 .t g(x. We assume that f is a concave function of x and that g is a convex function of x so that the first order conditions are not only necessary but sufficient for maximization. The Envelope Theorem – p.

θ) = 0 ∂x ∂L (x. θ) An interior solution of (3) where the constraint is binding satisfies the following first order conditions: ∂L (x. θ) − λg(x.The Lagrangian The Lagrangian is the following: L(x. θ) = f (x. 8/1 . θ) = 0. ∂θ (4) The Envelope Theorem – p.

θ) = λ (x. 9/1 . The Envelope Theorem – p. θ) = 0 (5) (6) The solution to this equation is the maximizer we are looking for. It is clearly a function of θ. Let’s denote it by x∗ (θ). θ) ∂x ∂x g(x.First Order Conditions Equivalently ∂f ∂g (x.

θ) = 0 ∂x ∂θ ∂θ or ∂x∗ ∂g ∗ ∂g ∗ (x (θ).Implications Note that by plugging x∗ (θ) into the constraint function g. Taking derivatives we get ∂g ∗ ∂g ∗ ∂x∗ (x (θ). θ) ≡ 0. θ) ∂x ∂θ ∂θ (7) The Envelope Theorem – p. 10/1 . θ) (θ) + (x (θ). we get the following identity (see (6)): g(x∗ (θ). θ) (θ) = − (x (θ).

We are interested in what happens to the optimal value of f when the parameter θ changes. V (θ) ≡ f (x∗ (θ). The Envelope Theorem – p. θ). Formally. dθ Note that V will change both because θ affects f and because it also affect the optimal choice of x. 11/1 .The question Let’s plug this optimal value into the objective function and define the optimal value of the objective function as a function of θ. we are interested in dV (θ).

θ) (θ) dθ ∂θ ∂x ∂θ ∂g λ ∂x (x∗ (θ). θ) (θ) = ∂θ ∂x ∂θ −λ ∂g (x∗ (θ). θ) + λ (x (θ).θ) ∂g ∗ ∂x∗ ∂f ∗ (x (θ). θ) − λ (x (θ). 12/1 . θ) = ∂θ where the second equality follows from (5).The answer To answer this question we take derivatives: ∂f ∗ ∂x∗ dV ∂f ∗ (θ) = (x (θ). The Envelope Theorem – p. and the last one from (4). θ) = ∂θ ∂θ ∂L ∗ (x (θ).θ) ∂θ ∂f ∗ ∂g ∗ (x (θ). the third one from (7). θ) + (x (θ).

The Envelope Theorem This proves the envelope theorem: the total rate of change in the optimal value of the objective function due to a small change in the parameter θ is simply the rate of change in the Lagrangian L evaluated at the optimal value of x. θ) dθ ∂θ The Envelope Theorem – p. 13/1 . dV ∂L ∗ (θ) = (x (θ).