Consider the effect of trade on income using the following model
yi =β0 +β1xi +β2wi +ei, where
yi is the log of income per capita,
xi is international trade,
wi is within-country trade,
and ei are shocks.
The shocks are likely to be correlated with the trade variables and propose to instrument xi using the country's proximity to other country's
xi = γ0 +γ1pi +vxi
and wi using country's size si xi using the country's proximity to other country's
wi = δ0 +δ1si +vw,i
assuming that E(vw,isi) = E(vx,ipi) = 0.
(i) Discuss the proposed identification strategy assuming that you observe a yi, xi, wi, pi, and si. What are the critical assumptions in these cross-country regression? Do you think they are plausible? Consider the effect of trade on income using the following model
yi =β0 +β1xi +β2wi +ei, where
yi is the log of income per capita,
xi is international trade,
wi is within-country trade,
and ei are shocks.
The shocks are likely to be correlated with the trade variables and propose to instrument xi using the country's proximity to other country's
xi = γ0 +γ1pi +vxi
and wi using country's size si xi using the country's proximity to other country's
wi = δ0 +δ1si +vw,i
assuming that E(vw,isi) = E(vx,ipi) = 0.
(i) Discuss the proposed identification strategy assuming that you observe a yi, xi, wi, pi, and si. What are the critical assumptions in these cross-country regression? Do you think they are plausible?