I have the following constrained maximization problem, written as a Lagrangian: $$ L(x,y,\lambda) = f(x,y) - \lambda(g(x,y)) $$ I can derive a set of implicit equations that characterize the solution, however, no closed form solution exists. I was thinking that I could apply the implicit function theorem to at least sign a set of partial derivatives (for example, $\partial x^* / \partial \alpha $, where $\alpha$ is some exogenous parameter).

Say I have, $$ \frac{\partial L(.)}{\partial x} = f_x - \lambda g_x = 0 $$ $$ \frac{\partial L(.)}{\partial y} = f_y - \lambda g_y = 0 $$ $$ \frac{\partial L(.)}{\partial \lambda} = g(x,y) = 0 $$ I can eliminate $\lambda$ and derive an implicit equation, $$ \frac{f_x}{g_x} - \frac{f_y}{g_y} = 0 $$ Which I combine with the budget constraint to get: $$ H(x,y) = \frac{f_x}{g_x} - \frac{f_y}{g_y} + g(x,y) = 0 $$ Can I then apply the implicit function theorem to $H(x,y)$ to evaluate the sign of $\partial x / \partial \alpha $ (where $\alpha$ is an exogenous parameter) evaluated at the optimal $x^*$?

Thanks for your help.


I'm aware that I could go the long route and set up a system of first order conditions to be solved using Cramer's rule. This would look something like, let $ H_1 = \frac{\partial L(.)}{\partial x}$ , $H_2 = \frac{\partial L(.)}{\partial y}$ and $H_3 = \frac{\partial L(.)}{\partial \lambda}$. At the optimal $x^*$, $y^*$, $\lambda^*$ each of these equations are equal to zero. Given an exogenous parameter $\alpha$ (in both $f(x,y$ and $g(x,y)$), what I'm after is (an element in) the solution to the following equation: $$\left( \begin{matrix} \frac{\partial H_1}{\partial x} & \frac{\partial H_1}{\partial y} & \frac{\partial H_1}{\partial \lambda} \\ \frac{\partial H_2}{\partial x} & \frac{\partial H_2}{\partial y} & \frac{\partial H_1}{\partial \lambda} \\ \frac{\partial H_3}{\partial x} & \frac{\partial H_3}{\partial y} & \frac{\partial H_3}{\partial \lambda} \end{matrix} \right) \left( \begin{matrix} \frac{\partial x^*}{\partial \alpha} \\ \frac{\partial y^*}{\partial \alpha} \\ \frac{\partial \lambda^*}{\partial \alpha} \end{matrix} \right) = - \left( \begin{matrix} \frac{\partial H_1}{\partial \alpha} \\ \frac{\partial H_2}{\partial \alpha} \\ \frac{\partial H_3}{\partial \alpha} \end{matrix} \right) $$ I was just hoping there would be an easier way to extract the information since $f(x,y)$ and $g(x,y)$ are very complicated functions. Thanks again.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.