Estimation for varying coefficient panel data model with cross-sectional dependence (Q2175225)

Consider a varying coefficient panel data model of the form \[ Y_{it}=\lambda_{i}+ \mathbf{X}_{it}^{\tau} \mathbf{m}\left(z_{t} \right)+U_{it}, \quad i=1,\ldots,N,\quad t=1,\ldots,T, \] where \(\tau\) denotes the transpose of a vector, \(Y_{it}\) denotes a one-dimensional dependent variable, \(\lambda_{i}\) is an individual fixed effect of individual \(i\), where \(\sum_{i=1}^{N} \lambda_{i}=0\), \(\mathbf{X}_{it}=\left(X_{it}^{\left(1 \right) },\ldots,X_{it}^{\left(p \right)} \right)^{\tau} \) is a \(p\times1\) vector, \(\mathbf{m}(\cdot) \) is a \(p\)-dimensional unknown varying coefficient function vector, \(z_{t}\) is one-dimensional covariate, \(\left\lbrace z_{t},U_{it} \right\rbrace \) are jointly stationary, and \(\left\lbrace U_{it}\right\rbrace \) are conditionally correlated and heteroscedastic. In this paper, the authors propose a consistent and efficient estimator for varying coefficient function vector \(\mathbf{m}(\cdot)\). From the authors' abstract: We first eliminate fixed effects by taking the cross-sectional average, and then use a local linear approach to obtain the initial estimator of the unknown coefficient functions. However, the initial estimator ignores the cross-sectional dependence and heteroscedasticity, which will lead to a loss of efficiency. Thus, we propose a weighted local linear method to obtain a more efficient estimator. In the theoretical part of the paper, we derive the asymptotic theory of the resulting estimator. Simulation results and a real data analysis are provided to illustrate the finite sample performance of the proposed method.