Studies have shown a decreasing trend in the U.S. gender earnings gap since the 1980s. The authors work with a framework established by Blau and Kahn (1997 JOLE; 2006 ILRR), who used the Michigan Panel of Income Dynamics (PSID) to decompose that gap into observable and unobservable components in order to determine which contributing factors are gender-specific and which can be attributed to the wage structure. They extend the Blau and Kahn framework to consider measurement error due to the use of proxy/representative respondents of the survey’s earnings variable. First, they find a trend toward more females in the gender composition of the household respondents, and second, they estimate the impact of that change on Blau and Kahn’s decomposition. They determine that some of the changes in the gender earnings gap could actually be attributed to whether the surverys were self- or proxy-responded. That is, the actual reduction in the gender pay gap may be smaller than what the estimates—without taking into account the measurement error—might indicate. The authors suggest the need for a careful validation study to ascertain the extent of the spurious measurement error effects.
Centre for microdata methods and practice