Estimation of Reporting Tax errors for Chinese Personal Income Based on Errors-In-Variables Model

This paper proposes the methodology of weighted nonparametric estimation of the conditional density to estimate real per capita annual disposable income in the year of 1995, 1998, 2001 and 2004 in China, and then investigates the effect of errors-in-variables due to intentional reporting errors for tax purpose, or tax evasion and evaluates a potential solution for bias due to errors-in-variables. The estimating results suggest that: First, there are intentional reporting errors of personal income or tax evasion. Intentional reporting errors accounted for over 21 percent of the corrected per capita annual disposable income in 1998. Second, intentional reporting errors constituted 16% of the total errors on average over the observed year, which was reflecting the seriousness of evading personal income tax.