Autor Dorn and Hanson 2013 — The China Syndrome Local Labor Market Effects of Import Competition in the United States

Summary

The foundational paper establishing the causal effects of Chinese import competition on U.S. local labor markets. Using a Bartik-style instrumental variables strategy — exploiting cross-commuting zone (CZ) variation in initial industry mix, instrumented by contemporaneous Chinese export growth to eight other high-income countries — the paper shows that CZs more exposed to Chinese import penetration suffered substantially larger declines in manufacturing employment, earnings, and overall employment-to-population, with workers exiting into unemployment, not in labor force (NILF), and transfer programs rather than reemployment in unexposed sectors. The paper's headline finding — no significant compensating employment gains in non-manufacturing, plus a $\$57.73$ /capita increase in annual government transfers per $\$1{,}000$ of import exposure — refuted the pre-2000 consensus that trade effects diffuse quickly through factor markets and that displaced workers readily reabsorb.

Key Claims

Instrument validity: Chinese export growth to Australia, Denmark, Finland, Germany, Japan, New Zealand, Spain, and Switzerland correlates at $0.92$ with U.S. industry-level import growth; the common driver is Chinese productivity and trade cost improvements (supply-side, internal to China), not correlated demand shocks to U.S. CZs.
CZ-level employment effects: Per $\$1{,}000$ increase in per-worker import exposure: manufacturing employment share $-0.60$ percentage points (pp), non-manufacturing employment $+0.00$ (insignificant), unemployment $+0.22$ pp, NILF $+0.55$ pp, log weekly wage $-0.76$ log pts, annual income per adult $-\$549$ .
No offsetting reallocation: Overall employment-to-population falls at least one-for-one with the manufacturing decline. No significant employment gains in unexposed sectors over a decade-long window.
Transfer cascade: Annual government transfers per capita rise $\$57.73$ per $\$1{,}000$ of exposure (1990–2007), dominated by Medicaid/Medicare ( $+\$18.27$ ), income assistance ( $+\$15.04$ ), Social Security Administration (SSA) retirement ( $+\$10.00$ ), Social Security Disability Insurance (SSDI) ( $+\$8.40$ ), and unemployment insurance (UI)/Trade Adjustment Assistance (TAA) ( $+\$3.65$ ). TAA — the designated trade adjustment program — is the smallest category; SSDI absorbs more than twice as much.
Trade deficit mechanism: The U.S. ran a sustained trade deficit with China throughout the period (U.S. deficit averaged $\approx 4.4\%$ of gross domestic product (GDP) in the 2000s); under balanced trade, displaced tradables workers would reallocate to unexposed tradables, but imbalanced trade channels them into nontradables or out of the labor force.
Industry-level evidence (Acemoglu et al. 2016 extension): $1$ pp increase in import penetration $\to$ $-1.30$ log pts manufacturing employment (two-stage least squares (2SLS)); total economy job losses $2.0$ – $2.4$ million including input-output linkages and CZ demand multiplier.

Concepts Introduced or Extended

Entities Mentioned

Quotes

"A $1,000 increase in import exposure per worker reduces employment and earnings and raises transfer payments substantially in commuting zones more exposed to Chinese imports."

"Trade Adjustment Assistance (TAA) — the program specifically designed to help workers displaced by trade — accounts for the smallest share of the transfer increase."

"We find no evidence of offsetting employment gains in non-manufacturing. Our estimates imply that exposure to Chinese import competition explains one-quarter of the contemporaneous aggregate decline in U.S. manufacturing employment."

My Take

The paper is arguably the single most influential causal identification of trade's labor market effects, and the Bartik-IV design has been widely replicated and extended. Three aspects deserve particular emphasis for this wiki's disability insurance (DI) focus:

The SSDI finding is underappreciated: Table 7's transfer decomposition shows SSDI absorbs more than twice the transfer increase of UI+TAA combined. This directly links the China trade shock to Conditional DI Applicants and to DI Growth Decomposition — displaced manufacturing workers use SSDI as permanent income replacement, not UI as temporary bridge. The mechanism is: job loss → long-term nonemployment → deteriorating health and rising DI applications, exactly the sequence documented in Autor and Duggan (2003).
Internal validity is strong; external validity requires caution: The exclusion restriction is plausible but depends on the 8 comparison countries' demand shocks being uncorrelated with U.S. CZ-level shocks. To the extent Europe and Japan experienced correlated demand contractions in the same period, the instrument is partially contaminated. Subsequent robustness checks and replications have been reassuring, but the concern is not zero.
No reallocation is the key empirical finding: The pre-2000 consensus treated rapid factor-market adjustment as axiomatic; Autor, Dorn, and Hanson (ADH) 2013 shows it doesn't happen on a decade timescale, especially for low-wage workers (see Autor, Dorn, Hanson, Song 2014 for worker-level tracking). This "stuck workers" finding is the mechanism behind both the persistent wage effects documented in Occupational Polarization and the DI take-up documented in Conditional DI Applicants.