Topcodes and the Great U-Turn in Nonmetro/Metro Wage and Salary Inequality

Part of the perceived increase in wage and salary inequality in the early 1980's may be due to social scientists using Bureau of the Census topcodes in Current Population Survey (CPS) data as if they were valid incomes. A topcode is the number that the Bureau of the Census substitutes for a reported income bigger than the maximum disclosable income in CPS public use sample files. Large incomes are rare and, consequently if disclosed, might allow the respondent to be identified, thus breaching the pledge of confidentiality from the Bureau of the Census to the respondent. In the 1960's, 1970's, and 1980's, the Bureau of the Census used the maximum disclosable income itself as the topcode. Estimating a measure of inequality using a topcode as if it were a valid observation on an income yields an underestimate. This downward bias was acute in the late 1970's when the Bureau of the Census did not raise its maximum disclosable income in a time of rapid inflation. It did though in the early 1980's, making some of the measured increase of income inequality in the early 1980's artificial. This downward bias on estimates of inequality in the late 1970's affected metro income inequality more than nonmetro because metro areas have a higher proportion of large incomes. Nonmetro inequality is historically higher than metro inequality. The nonmetro/metro gap in inequality was overestimated in the late 1970's.