Why isn't the VEP turnout rate the same as the Census Bureau's CPS turnout rate?

A frequently cited-source for turnout and voter registration rates is the Current Population Survey's biannual reports on voting and registration (aka CPS). The CPS is a large survey conducted by the federal government with the primary purpose of computing the unemployment rate. In the November of a federal election year, a limited supplement of voting and registration questions are asked. These data are among the best information for turnout and registration among various demographic subgroups -- such as the young or by race -- which cannot be computed from aggregate data presented here.

However, there are notable distinctions between the CPS turnout rate and the VEP turnout rate.

First, the two do not have the same denominator. The CPS sample frame is the resident non-institutional population of the United States. The VEP is broader in that it is an estimate of all persons eligible to vote, regardless if they live in an institution or overseas.

Second, the CPS is a survey, and is therefore susceptible to all survey methodology issues -- both from random sampling and non-random sampling issues. (Indeed, myself and others have questioned the dubious CPS practice of counting all persons who are never asked the voting question as having not voted.) Thus, even if the CPS and VEP had effectively the same denominator, these survey issues would still produce different estimates.