???????? ??? ???? ??? ??? ? ??? ? ????: 2013~2016? ??? ???? ??????? ?? (How well can we estimate the distribution of wage and salary income from the Statistical Yearbook of National Tax?: Comparison with 2013-2016 National Tax Service¡¯s thousand percentile data on wage an)

??? ?????????? ? ??? ???? ???? ?????? ???? ??? ??? ???? ??? ?? ???[Generalized Pareto Curves(GPC) Interpolation]?? ??? ???? ??? ??? ??? ??? 2013~2016? ?? ? ???? ?????? ????. ? ??? ?? ¡®???? ?????¡¯? ??? ? ?? ¡®???? ????¡¯? ????? ??? ???? ??? ????. ?? ? ???, ????????? ?????(????? ?? ?)? ???? 15?? ?? ????, ?????(????? ?? ?)? ???? 11?? ?????? ??? ?? ??? ?? ??? ???? ???, GPC ???? ??? ?? ??? ??? ?????? ?? ???? ??? ??? ??? ??? ? ??? ????. 4?? ?? ??? ?? 50%, ?? 40%, ?? 10%, ?? 1%? ?????? ¡À0.1%p ? ?? ??? ??? ? ? ?? ???. ????? 0.001 ?? 0.002 ?? ???? ? ??? ?????, ? ??? ??? ?????. ??? 0.1% ??? ??? ? ? 1% ? ????? ????? ?????? ?????? ??? ? ??, ??? ?? ?? 1% ? ??? ?????? 0.01~0.04%p ?? ??????? ??? ?????. ???? ??? ??? ?? ? ??? ??, ???? ??? ?? ??? ? ???? ???? ??? ??????? ???? ? ? ?? ?? ??? ??? ????? ???? ????? ?, ? ??? ???? ??? ???? ??? ?? ???? ?? ?? ???? ????? ?? ? ??? ??? ????.

This paper compares the estimated distribution of taxable wage and salary income, using the year-end settlement of taxable wage and salary income tabulation data presented by the Statistical Yearbook of National Tax, and Generalized Pareto Curves(GPC) interpolation, with 2013~2016 National Tax Service¡¯s thousand percentile data on taxable wage and salary income released by the politician Sang-Jung Sim. Both data have the same population and income source. Although the Statistical Yearbook of National Tax provides the very limited information on the whole distribution throughout these years, we can estimate the actual income distribution fairly accurately by using the GPC interpolation method. There has never been an estimate of more than ¡À0.1%p differences in the share of the bottom 50%, middle 40%, top 10% and top 1% over four years. The Gini coefficient is underestimated by 0.001 or 0.002, but the difference is negligible. However, when comparing the estimated distribution and the thousand percentile distribution in the top 1% group divided by 0.1%, it is found that the estimated distribution underestimates the income share of the top 1% bracket divided by 0.1% by about 0.01~0.04%p. The main contribu- tion of this paper is to identify the excellence of the GPC interpolation, to call for the need for statistical data presented in more income categories for the top income bracket, and to raise the need for regular and public disclosure of NTS¡¯s thousand percentile data.