Space Can Be Added To The Model Estimates And Inspection

As a branch of econometrics, spatial econometrics attracts more and more attention in recent years and has been studied thoroughly both in theory and application. Spatial regression, which is represented by space lag regression model and space residual autocorrelation model, is the basis of spatial econometrics, and also the most widely useful method in modeling. However, this classical spatial regression model belongs to parameter model. In data analysis in practice, specifying an incorrect parametric form can lead to serious errors in inference. Over the last two decades, some useful semiparametric models have been proposed to capture the underlying relationships between dependent variables and their associated covariates. Nonparametric and semiparametric models have been pain attention both in the statistical and econometrics fields. However, results on spatial data analysis by the nonparametric and semiparametric methods are few.Geoadditive model as a kind of semiparametric model can depict spatial effect flexibly, but its estimation method and testing method of inspection have not been studied in-depth. This paper firstly constructs the estimator of non-parameter and parameter parts based on the local linear method and profile least square method, then studies in the estimation and testing methods in the situation that there is constraint condition in parameter part. On the other hand, we conduct testing to prove that spatial effect affects dependent variables.