Input-Output as a Simple Econometric Model: Reply
暂无分享,去创建一个
variables-precisely the case examined by Gerkinghe finds that moments of the finite-sample distribution for the TSLS estimator exist only up to the number of overidentifying restrictions. In the context of equations (1) all structural equations are exactly identified. It follows that none of the moments of this distribution exist. One may obtain parameter estimates, but associated tests of significance are simply not meaningful. The empirical results established by Gerking must be questioned on these grounds. It should also be recognized that any estimator used to obtain structural coefficients in this model must ensure that both the input and the output identities are satisfied. When coefficient estimates are obtained they must be such that implied interindustry flows (Z13 = a^X1) are consistent with the equality of gross output and gross outlay. Without this constraint, comparative static results based on input-output coefficients are not meaningful.