Ja. Bishop et al., TRUNCATION BIAS AND THE ORDINAL EVALUATION OF INCOME INEQUALITY, Journal of business & economic statistics, 12(1), 1994, pp. 123-127
Lorenz dominance analysis is used to examine the effect of top-coding
on the ordinal evaluation of U.S. income inequality across time. Curre
nt Population Survey microdata are adjusted for truncation bias, and s
tatistical inference procedures are used to examine biennial changes i
n unadjusted and adjusted Lorenz curves. Beginning in 1985, the trunca
tion bias has a significant effect on ordinal rankings of income inequ
ality.