Confused by informal math: total differentiation

44 Views Asked by At

I'm reading these notes that say: total differentiation gives $$ P=a_LW+a_KR\implies dP=a_LdW+a_KdR+[Wd(a_L)+Rd(a_K)]\tag{i}. $$ Please let me explain the notation: we can think of $R,W$ and $P$ as rent, wage, and price respectively. The quantities $a_L$ and $a_K$ come from the minimization problem: $$ \min_{a_L,a_K} a_LW+a_KR\quad\text{s.t.}\quad a_K=g(a_L).\tag{ii} $$ The function $g$ doesn't depend on $W$ and $R$ and we know that the optimal $a_L$ and $a_K$ are interior solutions that only depend on $\frac{W}{R}$. Then, the author claims $[Wd(a_L)+Rd(a_K)]=0$ in (i) and so $$ dP=a_LdW+a_KdR.\tag{iii} $$ I can see why $[Wd(a_L)+Rd(a_K)]=0$ because it's just the first-order condition for (ii): $$ 0=W+g'(a_L)R\implies -\frac{W}{R}=g'(a_L)=\left.\frac{da_K}{da_L}\right|_{a_L=a_L(W/R)}. $$

My question: I can sort of follow why the author gets (i) and (iii) but I am uncomfortable by the seeming lack of rigor. What is and how to arrive at the rigorous analog of (iii) (e.g., $-\frac{W}{R}=\frac{da_K}{da_L}$ is the rigorous version of $[Wd(a_L)+Rd(a_K)]=0$)?

p.s. I would also appreciate some references so that I can understand things like this better in the future. Thank you!

Edit: I figured it out! I'll leave this post here for my own record and just in case someone else might find it useful.

1

There are 1 best solutions below

0
On

Typing things all out to MSE really helped!

The rigorous version of (iii) is $$ 1=a_L\frac{dW}{dP}+a_L\frac{dR}{dP}\tag{iv}. $$ To get here, we differentiate $$ P=Wa_L(W/R)+Ra_K(W/R) $$ w.r.t to $P$ to arrive at $$ 1=a_L\frac{dW}{dP}+a_L\frac{dR}{dP}+\left(\frac{d(W/R)}{dP}\right)\left[Wa'_L(W/R)+Ra'_K(W/R)\right]. $$ So we are done if the expression inside $[\cdot]$ is $0$, but this follows because $$ Wa'_L(W/R)+Ra'_K(W/R)=a'_L(W/R)[W+Rg'(a_L(W/R))]=0. $$