Borjas's timing rebuttal: the CPS Black-share jump was concentrated in one year (1979), while Miami's steepest low-skill wage drop came later (1981-1986), when the share was flat

Against Clemens and Hunt's census-artifact argument, George Borjas's direct reply argues the timing does not fit a purely compositional explanation. He reports the Black share of Miami's low-skill March CPS sample jumped from 36.3% in 1978 to 63.0% in 1979 — a single-year, single-survey-transition move around the 1980 census — and then stayed essentially flat from 1979 through 1983. But the steepest measured decline in Miami's low-skill wage occurred precisely in that flat window (and continued through 1986), not in the year the composition shifted. A parallel CPS-ORG (May/Outgoing Rotation Group) extract shows the same decoupling: wages fell sharply between 1980 and 1984 while the Black share moved only slightly. Borjas also cites the 1980 decennial census itself — taken weeks before the Boatlift — as already showing a 54.5% Black share in the comparable low-skill sample, consistent with his reading that the March CPS's sudden jump reflects a correction of prior undercounting rather than a Boatlift-timed artifact.

As a rebuttal to the simplest version of the compositional-artifact story — that the whole wage drop is explained by sampling composition — this is a specific, quantified objection: artifact and effect do not move together in time. It does not, on its own, settle the broader dispute; see claim-mariel-wage-effect-depends-on-race-adjustment-model-choice for the round that followed.