I am trying to get a good understanding of the steps involved in solving the dual of a maximization problem, namely cost minimization. At some point (last two steps), the author ends up with the following function for marginal cost growth: $$\Delta mc=\left [\frac{WN}{C(\cdot )} \right ]\Delta w+\left [1-\frac{WN}{C(\cdot )} \right ]\Delta r -\Delta e$$ Where the relation between price and marginal cost is given by: $$(1-B)P=MC=\frac{G(W,R)}{E}$$ From this latter equation, the difference between the change in price and a weighted average of changes in factor prices, the dual or price-based Solow residual is defined as: $$\alpha \Delta w +(1-\alpha)\Delta r-\Delta p=-B(\Delta p-\Delta r)+(1-B)\Delta e$$ Where the weights for the factor prices are the wage share in output for wages and its complement for capital costs ($\alpha$) and ($1-\alpha$).

I can't figure out how you get to this last step. I tried differentiating the marginal cost equation and substituting in the second one, but I'm missing some terms. The problem comes from the following paper: "Can Imperfect Competition Explain the Difference between Primal and Dual Productivity Measures? Estimates for U.S. Manufacturing" (1995).

*$\Delta x$ is the log difference of the variable X.


1 Answer 1


Equation (4b) in the paper gives: $$ \Delta mc_t = \frac{E_t N_t W_t}{Y_t G(.)}\Delta w_t + \frac{E_t K_t R_t}{Y_t G(.)} \Delta r_t - \Delta e_t. \tag{I} $$ Next we also have equation (5) in the paper: $$ (1-B)P_t = MC_T = \frac{G(.)}{E_t}. \tag{II} $$ Now substitute $(II)$ into $(I)$ to get: $$ \Delta mc_t = \frac{N_t W_t}{P_t Y_t (1-B)}\Delta w_t + \frac{K_t R_t}{P_t Y_t (1-B)} \Delta r_t - \Delta e_t. \tag{III} $$ Now, we define the wage cost over the total revenue to be $\alpha_t$. $$ \alpha_t = \frac{N_t W_t}{P_t Y_t}. \tag{IV} $$ Also, by the CRS assumption, we have that $\frac{N_t W_t}{C(.)} + \frac{R_t K_t}{C(.)} = 1$. Dividing this by $P_t Y_t/C(.)$ and using $C(.) = G(.)Y_t/E_t$ as equation (2) in the paper, we get: $$ \frac{N_t W_t}{P_t Y_t} + \frac{R_t K_t}{P_t Y_t} = \frac{C(.)}{P_t Y_t} = \frac{G(.)Y_t}{E_t P_t Y_t} = \frac{G(.)}{E_t P_t} = (1-B) \tag{V} $$ The last equality comes from comes from $(II)$ Using $(IV)$ in $(V)$ gives: $$ \frac{R_t K_t}{P_t Y_t} = (1-B) - \alpha_t. \tag{VI} $$ Now substitute $(IV)$ and $(VI)$ into $(III)$ to get: $$ \Delta mc_t = \frac{\alpha_t}{(1-B)} \Delta w_t + \frac{1 - \alpha_t - B}{(1-B)}\Delta r_t - \Delta e_t. \tag{VII} $$ Also log differentiating the first part of $(II)$ gives $\Delta p_t = \Delta mc_t$ so substituting this in $(VII)$ (and multiplying both sides by $(1-B)$ finally produces: $$ \Delta p_t (1-B) = \alpha_t \Delta w_t + (1 - \alpha_t - B) \Delta r_t - \Delta e_t. \tag{VIII} $$ The expression that you are looking for is a simple rearrangement of $(VIII)$.


Your Answer

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

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