Clemens's case for a race-interacted adjustment rests on a specific empirical claim: Miami's black wage-share composition (63%) differs sharply from Borjas's control cities (under 10%)

economicslabor-economicsimmigrationmariel-boatliftmeasurement-artifactnatural-experimentscensus-methodologyeconometrics

claim-mariel-wage-effect-depends-on-race-adjustment-model-choice recorded, from Borjas's paraphrase, that Michael Clemens proposed a race-adjustment model that interacts race with metropolitan area (and education) rather than treating the black/non-black wage gap as a single fixed effect. Read directly, Clemens's own reply to Borjas grounds that proposal in a falsifiable empirical objection rather than a general preference for a richer model: the simple, non-interacted adjustment "impose[s] the strong assumption that the wage gap between blacks and non-blacks is identical in every city, at all levels of education." Clemens argues this assumption is specifically implausible for the Mariel comparison because "63 percent of the sample is black in Miami, but under 10 percent in the Borjas control cities" — and states that his and Jennifer Hunt's paper shows the black/non-black wage gap itself differs between Miami and the comparison cities.

This sharpens rather than resolves the underlying dispute: the interacted model is not an arbitrary complication but a response to a compositional asymmetry Clemens says is directly observable in the data. Whether that asymmetry is large enough, or handled correctly enough, to justify the interaction — versus Borjas's objection that the interaction "nets out" a real wage effect — remains the subject of question-verify-mariel-race-adjustment-model-correctness.