---
title: "The measured wage effect of the Mariel Boatlift flips with sample construction — no effect (Card) versus a 10–30% drop for dropouts (Borjas)"
type: "claim"
status: "seedling"
source_url: "https://www.bruegel.org/blog-post/mariel-boatlift-controversy"
source_title: "The Mariel Boatlift Controversy"
source_author: "Silvia Merler"
source_date: "2017-06-05T00:00:00.000Z"
source_quote: "very drastic"
source_tier: 2
provenance: "Promotion from 10-inbox/raw/2026-07-09-hop-synthetic-control-mariel-boatlift.md, 2026-07-11"
origin: "batch"
derived_from: "10-inbox/raw/2026-07-09-hop-synthetic-control-mariel-boatlift.md"
writer_model: "claude-opus-4-8"
date_created: "2026-07-11T00:00:00.000Z"
audit_status: "flagged [unverified-quant — the load-bearing figures (Card: no effect; Borjas: 10–30% dropout wage drop; Peri & Yasenov: 15–20 Miami observations per year) rest on a Tier-2 secondary retelling (Merler, Bruegel), not on the primary papers. Only the fragment 'very drastic' is preserved as a verbatim quote. Primary verification (Card 1990 ILRR, Borjas 2017 ILRR, Peri & Yasenov synthetic-control paper) is routed to [[question-verify-mariel-boatlift-wage-effect-primary-sources]]]. Keep seedling until the numbers are checked against primaries. | AUDIT 2026-09-11 (claude-fable-5-1, cross-model): the routed question sits in _answered/ on a 2026-08-07 seek_verify log that matched 'very drastic' against the Bruegel page only — the primaries the question asked for had not been read. Read now. (1) Bruegel re-fetched: all four figures present as the note carries them ('virtually no effect on the wages or unemployment rates of less-skilled workers'; 'The wage of high school dropouts in Miami dropped dramatically, by 10 to 30 percent'; 'These very drastic choices leave the sample in Miami as small as 15-20 observations per year'; '125,000'). (2) Card 1990, ILRR 43(2):245–257, author-hosted PDF davidcard.berkeley.edu/papers/mariel-impact.pdf (extract_pdf sha256 0302f87b50999df5ed2aaaa2015e31a06e95daff6989b01500a2d1e193949867, 14 pp., TLS verified): abstract EXACT — 'the Mariel influx appears to have had virtually no effect on the wages or unemployment rates of less-skilled workers, even among Cubans who had immigrated earlier'; 'some 125,000 Cuban immigrants arrived in Miami'. CONFIRMED at Tier 1. (3) Borjas, NBER WP 21588 (Sept 2015, rev. July 2016; ILRR 2017), nber.org/papers/w21588 abstract: 'At least 60 percent of the Marielitos were high school dropouts. … The wage of high school dropouts in Miami dropped dramatically, by 10 to 30 percent'. CONFIRMED at Tier 1. (4) Peri & Yasenov, NBER WP 21801 (Dec 2015, rev. June 2017), abstract at nber.org/papers/w21801 and full PDF (extract_pdf sha256 ee4ccbb64df86a2eac691f5ef51a8ab5f7e266b0d1494ef4becdc2ffd3579840, 57 pp., TLS verified, read end-to-end by a read-only subagent): synthetic-control rebuttal and small-sample critique CONFIRMED ('Using a sample of non-Cuban high-school dropouts we find no significant difference in the wages of workers in Miami relative to its control after 1980'; 'Borjas' (2017) sample belongs to the sub-groups with less than 20 observations'; 'between a quarter to a third of our preferred sample'), but 'drastic' 0 occurrences and no '15-20 observations per year' — the only '15-20' is 'differences in average wage across cities within 15-20%'. Those two phrases are Merler's. Body attribution corrected with promotion wording preserved inline. Net: the [unverified-quant] flag on the Card and Borjas figures is discharged at Tier 1; the P&Y leg is confirmed in substance with its wording re-attributed. Title, claim, source_quote, Tier 2 for the cited Bruegel page, status seedling unchanged — status is the queen's call; the numbers now clear the floor for promotion."
tags: ["economics","labor-economics","immigration","mariel-boatlift","natural-experiments","synthetic-control"]
verified_verbatim: "2026-08-07 — source_quote matched verbatim (normalized) against a direct fetch of source_url by seek_verify (no model involved)"
seek_code_commit: "3b23cae"
audits: ["2026-09-11 claude-fable-5-1"]
---


The 1980 Mariel Boatlift — roughly 125,000 Cuban refugees arriving in Miami over months — is the canonical natural experiment on how a sudden low-skill labor supply shock affects native wages. Serious economists working the same government data have reached flatly opposite headline conclusions, and the divergence tracks sample-construction choices rather than the event itself.

David Card's 1990 study found no measurable effect on less-skilled Miami wages. George Borjas (2015/2017) restricted the sample to high-school dropouts and reported a wage drop on the order of 10–30% for that subgroup. Giovanni Peri and Vasil Yasenov rebutted Borjas using the [[claim-synthetic-control-built-to-measure-basque-terrorism-gdp-cost|synthetic control method]] — built originally for Basque conflict economics — to construct a comparison "Miami," and argued that Borjas's result depended on sample restrictions that left the Miami sample far too small — in Peri and Yasenov's own words, Borjas's sample "belongs to the sub-groups with less than 20 observations" and "is between a quarter to a third of our preferred sample," and "by focusing on small sub-samples and matching the control group on a short pre-1979 series, as done in Borjas (2017), one can find large wage differences between Miami and control because of large measurement error." *(Promotion wording: "argued that Borjas's result depended on 'very drastic' sample restrictions, leaving only 15–20 Miami observations per year." Corrected 2026-09-11: "very drastic" and "15–20 observations per year" are Silvia Merler's phrasing at Bruegel, not Peri and Yasenov's — neither string occurs in the June 2017 revision of their NBER working paper; the only "15-20" there is a wage-difference percentage. Merler's count is a fair gloss of P&Y's "a quarter to a third" of a 60–80-per-year preferred sample, but it is hers.)*

The structure recurs elsewhere in the vault: it is the same move as [[claim-soviet-emigre-career-premium-was-structural-not-ability]], where holding ability constant and varying arrival year reverses a talent story into a market-structure story. Here one holds the event (the boatlift) constant and varies the *comparison-group construction*, and the sign of the conclusion flips. A separate line of attack argues the whole effect may be a measurement artifact rather than a real labor-market response — see [[claim-mariel-wage-effect-may-be-1980-census-methodology-artifact]]. Card's Mariel work later anchored his [[claim-card-2021-nobel-natural-experiments-labour-economics|2021 Nobel]].

> [!note] Seek's commentary:
> "Hold the event constant, vary the estimator" is the transferable template — a warning that a natural experiment's headline number is often an artifact of researcher degrees of freedom in defining the treated and control samples. — Seek