Summary
Companion to Duleep and Dowhan's research on immigrant earnings, this paper specifies how to modify the three standard microsimulation earnings approaches — human capital, past-is-prologue, and donor — to handle the foreign-born. It demonstrates that earnings growth rather than earnings levels is the appropriate sufficient statistic for immigrant earnings trajectories, and shows how entry earnings relative to native peers predict subsequent growth. The paper applies the modified Modeling Income in the Near Term (MINT) microsimulation to project Social Security outcomes for immigrant cohorts through 2020, revealing dramatic shifts in the foreign-born retiree population and a widening gap in benefit adequacy.
Key Claims
- The three standard microsimulation earnings approaches must each be modified for immigrants: stratified by nativity, year of entry (pre/post-1980), and source-region (developed vs. developing country).
- Earnings growth — not the earnings level — is the sufficient statistic for predicting immigrant retirement outcomes; entry earnings relative to native peers serve as the operative predictor.
- Emigration follows an exponential decay: y=d+ae−bx; approximately 87% of emigration occurs within the first 10 years of residence.
- Developing-country immigrants (men 15–39) have 10-year emigration rates of ~7.7% vs. ~33.2% for developed-country immigrants, a ~4× difference that reshapes the composition of long-term residents.
- Flow projections use OIS (Office of Immigration Statistics) administrative data supplemented by "beacons of change" — structural policy or demographic shifts that would invalidate past-is-prologue extrapolation.
- Undocumented immigrants are estimated via the residual method (census minus legal stock); linked to the Earnings Suspense File for wage-credit attribution; distinguished as transient vs. stayer populations.
- MINT policy-impact projections: foreign-born share of retirees rises from 10.7% (current) to 14.4% by 2020; Hispanic Average Indexed Monthly Earnings (AIME) falls ~17% below native baseline; Asian AIME falls ~25%; Old-Age (OA)-benefit-ineligible retirees rise from 4.3% to 10.3%.
Concepts Introduced or Extended
Entities Mentioned
Quotes
"Earnings growth, not earnings levels, is the key to predicting immigrant retirement outcomes."
My Take
The paper's core methodological contribution — using earnings growth as the sufficient statistic and entry earnings as the growth predictor — elegantly sidesteps the selection problem in immigrant level comparisons. The emigration parameterization is empirically grounded but relies on historical cohort flows that may not generalize to post-2006 entry cohorts with different policy environments. The MINT projections are striking: doubling of OA-ineligible retirees (4.3%→10.3%) implies a large uninsured population among the elderly foreign-born that standard Social Security analyses miss entirely.