On second-order Taylor's-series approximation and linear homogeneity (Q1080093)

A function \(\Phi\) : \(R^ 2_{++}\to R^ 1\) is said to have a second- order Taylor series approximation interpretation if there exist real constants \(\alpha_ i\) \((i=0,1,...,5)\), \(\alpha_ i\neq 0\) \((i=1,2,...,5)\) and continuous, strictly increasing functions \(f: R^ 1_+\to R^ 1\), \(g: R^ 1_+\to R^ 1\) with inverse \(f^{-1}\), such that \[ \Phi (x,y)=f^{-1}(\alpha_ 0+\alpha_ 1g(x)+\alpha_ 2g(y)+\alpha_ 3(g(x))^ 2+\alpha_ 4(g(y))^ 2+\alpha_ 5g(x)g(y)). \] The function \(\Phi\) : \(R^ 2_{++}\to R^ 1\) is said to be linearly homogeneous if \(\Phi (\lambda x,\lambda y)=\lambda \Phi (x,y),\) \(\lambda >0\). The linearly homogeneous quadratic mean of order p is the function \(\Psi (x,y)=(\alpha_ 3x^ p+\alpha_ 4y^ p+\alpha_ 5x^{3/2} y^{1/2})^{1/p},\) \(p\neq 0\), \(\alpha_ i\neq 0\) \((i=3,4,5)\). The linearly homogeneous translog function is \[ \ln \chi (x,y)=\alpha_ 0+\alpha_ 1 \ln x+\alpha_ 2 \ln y+\alpha_ 3(\ln x)^ 2+\alpha_ 4(\ln y)^ 2+\alpha_ 5 \ln x \ln y \] where \(\alpha_ 1+\alpha_ 2=1\), \(2\alpha_ 3+\alpha_ 5=0\) and \(2\alpha_ 4+\alpha_ 5=0\). The authors prove the following theorems: Theorem 1. Let \(\Phi\) : \(R^ 2_{++}\to R^ 1\) be linearly homogeneous. It is linearly homogeneous translog function if and only if it has a second-order Taylor's series approximation interpretation. Theorem 2. Let \(\Phi\) : \(R^ 2_{++}\to R^ 1\) be linearly homogeneous. It is linearly homogeneous quadratic mean of order p function if and only if it is a generalized quasiquadratic function.