Research of Decision Variables of Tax Revenue Based on Nonparametric Selection Approach - Innovative Management in Information and Production - page 303

Research of Decision Variables of Tax Revenue
Based on Nonparametric Selection Approach

Zuiyi Y. Shen and Bing Xu

Abstract To ensure sustainable tax revenue growth, decision variables of tax
revenue were researched in this paper by using mixed nonparametric kernel method
based on a dataset of tax revenue containing both continuous and discrete data in
Hangzhou; compared to parameters regression method, it can automatically reduce
the dimension of model and possess high model fitting precision. The mean square
error of mixed nonparametric kernel method reduces by 49.3 % after deleting 12
irrelevant variables, while it increases by line regression method and stepwise
regression. In addition, it is found that the variables of real-estate investment are not
a decision variable of tax revenue in nonparametric methods, which is different from
stepwise regression. This conclusion provides the negative evidence of experience
in tax revenues about whether it will produce decisive influence to tax revenues to
control and regulate the real-estate investment.

Keywords Tax revenue ￿ Decision variable selection ￿ Mixed nonparametric
kernel estimate

1 Introduction

The proportion of tax revenue to GDP reflects the relationship of possession and
control of social resources between the government and microeconomic subject.
It also reflects the degree and status of regulation of economic operation and