1
$\begingroup$

Suppose we levy a specific tax $t$, then, at the new equilibrium, we have $$ D(p+t)=S(p), $$ where $D$ is the demand function and $S$ is the supply function. And this makes equilibrium price $p$ is a function of $t$. The incidence formula $$ \frac{dp}{dt} = \frac{D'}{S'-D'}, $$ where $D'=dD(p)/dp$ and $S'=d S(p)/dp$, can be obtained through https://scholar.harvard.edu/files/stantcheva/files/lecture3.pdf.

However, some texts state that we can also obtain this formula through implicit differentiation with respect to $t$. My question is how we can do this.

This is my approach: \begin{aligned} &D(p+t)=S(p) \\ \implies &\frac{d D(p+t)}{d (p+t)} \frac{d (p+t)}{d t} = \frac{d S(p)}{d p)} \frac{d p}{d t} \\ \implies &D'(p+t)(\frac{d p}{d t}+1)=S'(p) \frac{d p}{d t} \end{aligned} It seems that only when $D'(p+t) = D'(p)$, I can obtain the same tax incidence formula. But is $D'(p+t) = D'(p)$ TRUE?

p.s. I know Is there an intuitive explanation for the tax incidence formula from elasticity? provides a good derivation way but it is still not through implicit differentiation.

$\endgroup$
1
  • $\begingroup$ Ah...I think it is certain that $\frac{d D(p+t)}{d (p+t)} = \frac{d D(u)}{d u}$... $\endgroup$
    – dchao
    Commented Oct 21 at 13:17

1 Answer 1

3
$\begingroup$

The implicit function theorem is handy here.

Rewrite the equilibrium condition as \begin{equation} F(p,t):=S(p)-D(p+t) = 0 \end{equation} Using the implicit function theorem, we have \begin{equation} \frac{\mathrm dp}{\mathrm dt} = -\frac{\partial F/\partial t}{\partial F/\partial p} = -\frac{-D'}{S'-D'} = \frac{D'}{S'-D'} \;, \end{equation} as desired.

$\endgroup$
8
  • $\begingroup$ Thanks. I have a question: is $\partial F / \partial t = 0 - D' \cdot 1$? If it is true, it seems that $p$ is independent of $t$ so that $d S(p) /d t = 0$? Why is not $d S(p) / dt = S' \cdot dp/dt$? $\endgroup$
    – dchao
    Commented Oct 21 at 15:10
  • 1
    $\begingroup$ But from another perspective, I derive the result. $0=dF = F_p dp + F_t dt$. That is, $0 = (S'-D')dp + (-D') dt$. Then, we get the desired $dp/dt$. $\endgroup$
    – dchao
    Commented Oct 21 at 15:13
  • $\begingroup$ And back to my question, in my approach, $D'(p+t)=D'(p)$ correct, right? $\endgroup$
    – dchao
    Commented Oct 21 at 15:16
  • 1
    $\begingroup$ @dchao: You have to careful with the prime notation of derivative. Let $x$ be the argument of the function $D$. Then, $$D'(p+t)=\frac{\mathrm dD(x)}{\mathrm dx}\Bigg\vert_{x=p+t}\quad\text{and}\quad D'(p)=\frac{\mathrm dD(x)}{\mathrm dx}\Bigg\vert_{x=p}.$$ The two are equal only if $t=0$. However, regardless of the value of $x$, $D'(x)$ is still same function defined as $\frac{\mathrm dD(x)}{\mathrm dx}$, which is probably what you meant by the equality of the derivatives. $\endgroup$
    – Herr K.
    Commented Oct 21 at 15:59
  • $\begingroup$ @ And in my first comment, do we treat $p$ and $t$ independent when we deal with the partial derivatives like $\partial F / \partial t$? $\endgroup$
    – dchao
    Commented Oct 21 at 16:04

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.