Applied Nonparametric Regression Analysis: the Choice of Generalized Additive Models

Literature has documented tremendous changes in classical regression analysis techniques since 1980s. The drawbacks of simple and multiple parametric regression analyses on model specifications and the non-robust assumption of error terms followed by the introduction of a series of diagnostic tests to fix these inevitable pitfalls have made econometricians to develop new methodologies in nonparametric and semi-parametric regressions that either do not have or mitigate the major shortcomings of what their traditional counterparts inherently demonstrate. The development of the generalized linear models followed by the introduction of generalized additive models and generalized additive mixed models has attracted practitioners to use these methodologies in applied studies. The main objective of this paper is to conduct a comprehensive survey on studies that used generalized additive models as econometric models and show how the parameters of these models are estimated. In particular, it briefly reviews the theory of generalized additive models, and then introduces various techniques to estimate the parameters of the models. Finally, it presents a comprehensive review of studies in which generalized additive models are specified as the econometric model.