Downstream methods and evidence

Scope and interpretation

Default results are illustrative reference calculations. They are conditional modeled expected differences, not forecasts about identified people, observed attributable counts, or identified policy effects. Target-cohort applicability requires structured demographics and a reviewed transport decision. Grandchild and later earnings outcomes are structural projections.

Worker displacement, net tradable-job loss, and changes in immigration exposure are distinct inputs. A displacement count does not establish either of the other exposures. The mortality response comes from high-tenure male workers exposed to Pennsylvania mass layoffs in the early 1980s. The selected baseline is US men aged 45–54 in 2015–2019. The child earnings estimate concerns Canadian father-son pairs after firm closure. Moving those estimates to another population, place or period requires an explicit transport rationale.

Each causal chain preserves citations, evidence roles and composition kinds. Relative earnings gaps transmit in gap space rather than by multiplying income levels. The same intergenerational earnings relationship is reused at each generation; it has one shared uncertainty draw. The implemented log_elasticity_transmission ensemble variant compares the finite-change v**t mapping with the default linear gap rule, using the same coefficient. Both are structural transmission assumptions, not independently identified intervention responses.

Count calculations

Mortality

Source-offset profile: displacement, +1, +2–3, +4–5, and +6+; displacement is follow-up year 1, so offset +6 begins in follow-up year 7. Annual baseline rates use a constant-hazard rate-to-risk conversion; odds ratios modify annual risk, and survival compounds across phases. The selected baseline age rate is held fixed over the follow-up horizon.

For annual event/person-year rate λ, annual risk m = 1 − exp(−λ). For odds ratio r, exposed annual risk q = r m / (1 − m + r m). Expected excess deaths = N × (counterfactual survival − exposed survival). Counterfactual survival is (1 − m) raised to the follow-up years; exposed survival is the product of (1 − q) raised to each phase duration. Deaths deplete the surviving cohort. Fractional durations assume constant hazards within a phase. The five phases cover offsets 0, +1, +2–3, +4–5 and +6 onward.

Demographic profiles must match the admitted male 45–54 all-cause response. Other baseline rates do not establish effect transport. An explicit profile takes precedence over county mortality. County rates require compatible sex, age, cause, calendar window and national prior, and a Gamma–Poisson posterior verified against deaths and person-years. Suppressed counties remain absent; national fallback is disclosed.

Unknown-covariance diagnostic

The separate additive-coefficient diagnostic uses marginal standard errors without inventing covariance. For one contrast T = sum(aᵢ βᵢ), let lᵢ = |aᵢ| SEᵢ. Its sharp standard-error bounds over compatible positive-semidefinite covariance matrices are max(0, 2 max(lᵢ) − sum(lᵢ)) and sum(lᵢ). A Gram-vector representation proves attainability by alignment or polygon closure. The point ± 1.96 times the upper bound is an outer normal-approximation envelope, not a source-reported interval or simultaneous band. Separate phase extrema need not be jointly attainable. This diagnostic does not replace the production mortality profile or create joint survival draws.

Child lifetime earnings

Undiscounted lifetime loss = displaced workers × children per worker × relative adult earnings gap × the cited US male synthetic career earnings baseline. This is a lifetime dollar amount in 2024 dollars, not an annual loss. The baseline sums cross-sectional age-band medians and approximates individual lifetime sums; it excludes nonworking years. The Canadian father-son effect and US dollar conversion each require a transport decision.

The source child cohort is boys aged 10–14 at paternal displacement. Known age/sex mismatches block the child-dollar count; unknown child profiles cannot qualify for reviewed headline use. Optional income-quartile contrasts retain every reported coefficient and its uncertainty, including null and beneficial possibilities. Quartiles refer to father's average pre-displacement income in 1978–1979 and require an explicit target-rank definition.

Optional child valuation requires known child ages, a real discount rate and a valuation year equal to the event year. Annual synthetic career earnings are discounted by (1+r) raised to years after that date, with dollars kept in 2024 purchasing power. The observed adult-age window (25–33) is shown separately from the constant-gap career projection (20–64). Neither recovers a source-identified annual earnings path; the career path extrapolates beyond observed ages. Zero-rate career valuation matches the undiscounted reference.

Receiving-community comparisons

The receiving-community command uses separate source-specific shares. Light–Miller applies a percentage-point change in undocumented population share to log violent-crime rates. Bianchi applies a log regular-resident share ratio to log total-crime rates. Both population denominators must be retained. Source windows and conditional fixed-effects versus IV roles remain distinct. Optional count differences use the same post population in both annual endpoint rate scenarios. No displaced-worker-to-arrival bridge, cumulative-crime extrapolation or additive crime total is supplied.

With reference/post shares s₀ and s₁, Light–Miller uses exp(100 β (s₁ − s₀)); Bianchi uses exp(β log(s₁/s₀)) and requires positive shares. For baseline b incidents per 100,000 and post population P₁, the endpoint difference is (b P₁ / 100000) × (rate ratio − 1). The arithmetic rate interpretation assumes an unchanged residual retransformation factor. Correct source units do not establish target transport.

Local service jobs

Local service jobs lost = T × ℓ, where T is independently documented net local loss of tradable jobs and ℓ is the multiplier for a declared job class. Manufacturing and high-tech multipliers represent different classes, not endpoints of one uncertainty interval. The tradable-share field is metadata and does not infer T from displaced workers.

Reference example

For 1,000 displaced workers, two children per worker and 20 years of mortality follow-up, the shipped national reference produces 28.76 expected excess deaths and $478,894,046.40 undiscounted child lifetime earnings loss. Both are illustrative and ineligible for public headline treatment until applicability is reviewed. Service-job losses remain blocked without a separate net-job input and job class.

Uncertainty and dependence

Low/high values are parameter-support envelopes, not confidence intervals. Monte Carlo p05–p95 is a central 90% interval for parameter uncertainty only. Exposure, baseline estimates, county measurements and pooling weights are held fixed; unmodeled pathways and structural uncertainty are excluded. Latin hypercube sampling preserves marginal coverage; declared correlations use Iman–Conover rank induction. Repeated relationship aliases share exactly one draw. Sobol attribution groups them, and dependent parameters are reported as joint blocks. Independent-step analytic moments refuse repeated shared variables. Predictive mortality, where available, adds binomial cohort variation in separately simulated exposed and counterfactual cohorts; their contrast is not an observable paired individual causal effect.

Per-result effect-removal and one-input source-endpoint experiments refit the same scenario while retaining all other inputs and admission gates. A neutral ratio is one; additive/job and place-coefficient removal uses zero where the outcome defines it. Removal changes need not sum because propagation is nonlinear, and removal can leave observed support. These are hypothetical comparisons, not causal shares. Endpoint spans are not variance indices, and fixed inputs acquire no inferred distribution.

A separate experimental uncertainty command samples a documented national event/person-time baseline using a Poisson likelihood and declared Jeffreys rate prior: Gamma(events + 0.5, rate = person-years). Baseline counts must reproduce the verified rate and match its sex, age, cause, calendar window and geography. Exposure is a caller-declared discrete uniform over integer bounds; it is a sensitivity assumption, not an inferred confidence distribution. Baseline, exposure and study parameters are assumed independent. Expected-effect intervals, independent-cohort process contrasts and the original fixed-input reference remain separate. The budget requires national source-profile mortality, excludes demographic-profile overrides, and does not sample county posteriors, transport error, age/time trends or probabilities over structural models.

Mortality persistence sensitivity

The default source-profile calculation also reports a declared alternative that retains the early odds ratios and sets source offset +6 onward to OR=1, first changing follow-up year seven. Both structures use the same resolved baseline, exposure and horizon. For the 1,000-worker, 20-year national reference, the source profile gives 28.76 expected excess deaths and the no-sustained-excess alternative gives 20.52.

The alternative is motivated by the long-run divergence recorded for Eliason–Storrie. It is a hybrid structural assumption, not a reproduction of the Swedish study or a substitution of its hazard ratios for Sullivan–von Wachter odds ratios. A nonsignificant estimate does not establish an exactly zero effect. The cutoff follows the source profile, not the Swedish study's windows; persistence beyond its 12-year follow-up remains an assumption.

Scenario Monte Carlo reports separate parameter-only intervals within each structure and a difference calculated on shared parameter draws. The range across their point estimates is not a confidence interval or a bound on the true effect; no probabilities or averaging weights are assigned to the structures. The source estimate remains unchanged. Both structures retain fixed-baseline aging and population-transport limits, and the sensitivity report is ineligible for a validated public headline.

Cohorts, aging and household comparisons

The pinned CPS January-2024 reference reproduces the published BLS known-year displacement total and includes separate three- and six-year-tenure cohorts. Interview age and household composition are not historical event characteristics, and these reference cohorts do not identify H-1B displacement.

The mortality-cohort command aggregates sourced male age-group contrasts exactly and compares aging age-band baselines with fixed baselines. Its response support is men aged 30–69 at mass layoff or firm closure with at least six years of tenure. It requires a declared constant projection of Table 7 source-window annual-logit responses, distinct from the default Table 5 phase profile. Event age selects the response, which stays fixed as survivors age through 2015–2019 period-rate bands. Uncovered workers remain unknown; partial coverage never becomes a whole-cohort total. These are Cartesian envelopes of marginal coefficient ranges, not simultaneous confidence bands. They exclude baseline variance, exposure error, cross-coefficient covariance and transport error, and are ineligible for public headlines.

The household command compares annual employment-state probabilities and gross earnings under a supplied unemployment duration. It calculates member marginal expectations exactly over one to five years, reusing observed SIPP 2021–2022 conditional transitions under stationarity and Markov assumptions, not an identified displacement recovery curve. Adults stay within ages 25–54; children age without an inferred earnings effect. Any reemployment earnings multipliers are declared assumptions. Sparse cells require explicit acceptance and no full household prediction interval is claimed; partner responses, joint household outcomes, taxes/transfers, savings, marriage, fertility and protective interventions remain unmodeled.

The uncertainty command adds documented national baseline count uncertainty, a declared integer exposure distribution and binomial variation to parameter draws. It states the Poisson/Gamma model and component-independence assumptions. Transport error and structural model probabilities remain excluded; parameter-only results and independent-cohort process contrasts retain separate labels.

Validation status

V0 checks internal consistency; it does not validate the model against an independent event. V1 historical transport diagnostics use the same five-phase mortality calculation as production, but population differences and fit failures remain visible. Independent-event V2 and prospective V3 validation have not been established. Code repairs and passing tests do not supply missing empirical validation.

Current mortality scorecard

Measured points lie inside the modeled bands for 0 of 3 mortality variants. These compare modeled mortality with a male–female differential; the population/outcome mismatch limits interpretation. Point coverage is neither a test against the measured confidence interval nor evidence of causal validity.

Excess deaths per 100,000 adults per unit shock
Variant	Model point [band]	Measured point	Covered?
all adults	18.78 [5.51, 42.87]	4.27	no
all male	27.17 [7.98, 61.94]	4.27	no
all male age matched	25.88 [7.6, 59.02]	4.27	no

The archived October 14 expanded evaluation reports a cohort reproduction difference of 128.3180 workers against the published 2,578,000-worker benchmark, within its 1,000-worker rounding tolerance. The SIPP baseline check includes 54 held-out calibration cells; the largest absolute transition-probability difference is 0.3759. There are 0 compatible pooled reports and 23 explicit pooling refusals. These are archived descriptive reproduction/baseline checks, not causal displacement validation or a new independent-event score.

The evidence selection protocol is locally frozen research work, not retrospective preregistration or an independently timestamped forecast. The archived results are retained in validation/followthrough_evaluation_expanded_20261014.json.

Consumer scaling and mixtures

Declared display factor s multiplies expected outcomes and corresponding experiments; output variance scales by s². Compatible mixtures use sum(wₖ E[Xₖ]) while retaining every baseline and applicability decision. Linearity of expectation does not require independence, but does not identify a joint distribution. Quantiles and variance indices cannot generally be averaged. A positive-weight unavailable component withholds the total. Joint mixture intervals remain unavailable until a joint draw/dependence calculation exists.

The code and admitted data define behavior. The paper explains the mathematics; software guides describe calling contracts and integration.

Parameter inventory

Download the complete parameter CSV with full identifiers, precision and notes.

Parameter support bands, destination units and evidence roles
Relationship and identifier	Point [low, high]	Unit	Evidence role	Sources, population and qualifications
displacement → worker earnings Full identifier `displacement->worker_earnings`	0.8 [0.75, 0.85]	gap_multiplier	conditional	jacobson1993;oreopoulos2008;davis2011 US high-tenure displaced men long-run earnings multiplier; band = cross-study spread: JLS early-80s PA losses vs D&vW 2011 full-text synthesis (still 15-20% below controls after 20y)
earnings shock → mortality sustained Full identifier `earnings_shock->mortality_sustained`	1.1354 [1.0335, 1.2474]	odds_ratio	conditional	sullivan2009 US high-seniority displaced men born 1930-1959 Sullivan & von Wachter NBER WP 13626 (2007), Table 5 column 3, offset +6 onward: log-odds 0.127 (SE 0.048), transformed as exp(coefficient +/- 1.96 SE) and rounded to parameter-file precision. Source offsets are relative to displacement; with displacement as follow-up year 1 this begins in year 7. The profile is sourced coherently from this working-paper table; reconcile against the published QJE version before treating the locator as published-version extraction. High-seniority male Pennsylvania mass-layoff cohort in the early-1980s recession; not automatically transportable to ordinary layoffs.; browningheinesen2012 CROSS-CHECK (v1.38): 20-year cumulative hazard ratio 1.10 [1.05, 1.16], still significant at 20y (authors rebut harvesting: yearly hazard ratios exceed 1 in almost all years) vs the SvW sustained annual OR 1.135 - different estimands and windows, not a like-for-like replication; cumulative hazard ratios cannot validate an annual OR magnitude or be pooled with it; bloemen2018 (v1.43): Dutch men 45-59, 5y +0.5968pp (SE 0.1763) = +33.5% (2y 60.9%, 3y 39.4%, 4y 41.2% - declining pattern consistent with peak-then-decay); DECLARED DIVERGENCE: eliasonstorrie2009 reports long-run all-cause estimates compatible with the null (5-8y HR 0.98, 9-12y 0.91, both n.s., 12y follow-up, baseline-displacement design) - the sustained-effect SIZE is design-heterogeneous across countries; positions never averaged (CITING 5), the SvW anchor stays; REPLACED 2026-10-14 v1.51: source beta/SE transformation re-executed at four decimals; prior stored point/low/high 1.135/1.033/1.247 remain in Git. Numerical source is the working-paper version; canonical publication tier under CITING section 1, pending published-version reconciliation.
earnings shock → mortality peak Full identifier `earnings_shock->mortality_peak`	2.6698 [2.1144, 3.3711]	odds_ratio	conditional	sullivan2009 US high-seniority displaced men born 1930-1959 Sullivan & von Wachter NBER WP 13626 (2007), Table 5 column 3, displacement offset: log-odds 0.982 (SE 0.119), transformed as exp(coefficient +/- 1.96 SE). The profile is sourced coherently from this working-paper table; reconcile against the published QJE version before treating the locator as published-version extraction. High-seniority male Pennsylvania mass-layoff cohort in the early-1980s recession; not automatically transportable to ordinary layoffs.; browningheinesen2012 CROSS-CHECK (v1.38): Danish plant closures (males 20-60, PSW), year-of-displacement cumulative hazard ratio 1.84 [1.44, 2.34] = +84% within its own study; the cumulative hazard ratio is not numerically comparable to the annual SvW OR without estimand/baseline harmonization; rechecked from 2011 AKF working-paper Table 6.1, PDF page 23; CROSS-COUNTRY FINDINGS WITH DIFFERENT SCALES AND WINDOWS (v1.43): eliasonstorrie2009 Swedish PSW establishment closures, men <=4y all-cause HR 1.44 [1.19, 1.76] (external 2.07 [1.42, 3.02], suicide 2.15 [1.28, 3.59], alcohol-related 2.21 [1.14, 4.31] beside); bloemen2018 Dutch LPM, men 45-59 at closing firms, year-1 +0.2229pp (SE 0.0812) = +85.8%; REPLACED 2026-10-14 v1.51: source beta/SE transformation re-executed at four decimals; prior stored point/low/high 2.6696/2.1151/3.3696 remain in Git. Numerical source is the working-paper version; canonical publication tier under CITING section 1, pending published-version reconciliation.
earnings shock → mortality offset 1 Full identifier `earnings_shock->mortality_offset_1`	1.9838 [1.3697, 2.8732]	odds_ratio	conditional	sullivan2009 US high-seniority displaced men born 1930-1959 Sullivan & von Wachter NBER WP 13626 (2007), Table 5 column 3, offset +1: log-odds 0.685 (SE 0.189), transformed as exp(coefficient +/- 1.96 SE). The profile is sourced coherently from this working-paper table; reconcile against the published QJE version before treating the locator as published-version extraction. High-seniority male Pennsylvania mass-layoff cohort in the early-1980s recession; not automatically transportable to ordinary layoffs.; REPLACED 2026-10-14 v1.51: source beta/SE transformation re-executed at four decimals; prior stored point/low/high 1.9838/1.3672/2.8777 remain in Git. Numerical source is the working-paper version; canonical publication tier under CITING section 1, pending published-version reconciliation.
earnings shock → mortality offsets 2 3 Full identifier `earnings_shock->mortality_offsets_2_3`	1.7315 [1.3238, 2.2649]	odds_ratio	conditional	sullivan2009 US high-seniority displaced men born 1930-1959 Sullivan & von Wachter NBER WP 13626 (2007), Table 5 column 3, offsets +2 through +3: log-odds 0.549 (SE 0.137), transformed as exp(coefficient +/- 1.96 SE). The profile is sourced coherently from this working-paper table; reconcile against the published QJE version before treating the locator as published-version extraction. High-seniority male Pennsylvania mass-layoff cohort in the early-1980s recession; not automatically transportable to ordinary layoffs.; REPLACED 2026-10-14 v1.51: source beta/SE transformation re-executed at four decimals; prior stored point/low/high 1.7316/1.3244/2.2641 remain in Git. Numerical source is the working-paper version; canonical publication tier under CITING section 1, pending published-version reconciliation.
earnings shock → mortality offsets 4 5 Full identifier `earnings_shock->mortality_offsets_4_5`	1.2712 [0.97, 1.6661]	odds_ratio	conditional	sullivan2009 US high-seniority displaced men born 1930-1959 Sullivan & von Wachter NBER WP 13626 (2007), Table 5 column 3, offsets +4 through +5: log-odds 0.240 (SE 0.138), transformed as exp(coefficient +/- 1.96 SE). v1.50 corrects the former SE 0.077 copied from column 6 offsets +2-3. Visually re-extracted from PDF page 54; source SHA256 20617cbe3924c8b58f6d3888cb0a0ea26ca908d60486f4aad304398dd6ef9c8f. Stable employment 1974-79 at employer over 50 workers; birth cohort 1930-59. The 45-54 baseline is a reference stratum, not the full source sample. Reconcile against the published QJE version before treating this as a published-version extraction; not automatically transportable to ordinary layoffs.; REPLACED 2026-10-14 v1.51: source beta/SE transformation re-executed at four decimals; prior stored point/low/high 1.2712/0.9700/1.6661 remain in Git. Numerical source is the working-paper version; canonical publication tier under CITING section 1, pending published-version reconciliation.
displacement → divorce hazard Full identifier `displacement->divorce_hazard`	1.11 [1.05, 1.25]	rate_ratio	conditional	rege2007;charles2004 married displaced workers Norway plant closures: marriages +11% more likely dissolved by 2003 (80k couples); Charles & Stephens 2004 confirm direction; band = cross-study spread widened for US generalization; internal evidence table said +23% - corrected 2026-09-06 against DP514 abstract; HALLA-SCHMIEDER-WEBER 2020 CROSS-CHECK (v1.43): Austrian registers, quarterly divorce hazard +0.004pp (SE 0.001) vs clean-firm controls = ~0.5pp cumulative over 5 years on a ~6% base (relative ~1.08 over 5y) - an order smaller than the shipped multiplier; PRECISE ZERO vs mass-layoff-firm controls (-0.001, SE 0.001) - firm-level uncertainty, not the displacement per se, carries the small effect; huttunenkellokumpi2016 beside (Finnish plant closures: divorce dpl1 +0.011 n.s. female, +0.011* male) - divergence declared, never averaged (CITING 5), rege2007 stays the anchor
displacement → child earnings Full identifier `displacement->child_earnings`	0.9076 [0.844, 0.976]	gap_multiplier	conditional	oreopoulos2008 Canada 39k father-son pairs firm closures REPLACED v1.11: was 0.91 [0.86,0.96] abstract-derived; Table 2 col 3 (income-controlled, N=38342) log coeff -0.097 (SE 0.037) -> multiplier exp(-0.097)=0.9076, 95% CI [0.844,0.976]; band WIDENED by evidence (CITING 4: bands never shrink); col 4 firm-controls -0.092 (.037) sits inside; huttunen2019 CROSS-CHECK (v1.37, recorded beside, NOT replacing: CITING 4 bands never shrink): Finland plant closures 1991-2000 (ILA matched employer-employee registers), child earnings at age 30 Table 2 col 3: -575.476 EUR (SE 278.543) on mean 25883.871 -> -2.2% (author prose), i.e. multiplier ~0.978 [1.96 SE band 0.957, 0.999]; males -2.4% (Table 6a: -762.501 SE 427.932 on 30647.315), females and displaced-mother households NULL (Table 6b/7); same-plant employment -0.006 (0.003) and same study choice -0.013 (0.006) at 25 -> operating channel is career/education choice, not skill formation (no GPA/crime effects, Table 4a). Nordic institutions + recession-cohort context; cross-jurisdiction spread vs the Canadian -9.2% anchor recorded honestly
divorce → child earnings Full identifier `divorce->child_earnings`	0.95 [0.9, 0.99]	gap_multiplier	boundary	gruber2004 US divorce-law-reform cohorts adult children of easier-divorce states: lower earnings + attainment, higher own divorce; PARALLEL stream (never composed with the displacement path silently)
family size → child earnings Full identifier `family_size->child_earnings`	0.98 [0.95, 1]	gap_multiplier	boundary	black2005 Norway admin per-child penalty for larger families / later birth order; PARALLEL stream
child earnings → grandchild earnings Full identifier `child_earnings->grandchild_earnings`	0.55 [0.4, 0.6]	gap_multiplier	structural	solon1992;corak2013;chetty2014 US IGE literature IGE band; composes in GAP SPACE: grandchild gap = 1 - IGE*(1 - child gap); BEHRMAN-TAUBMAN 1990 CROSS-CHECK (v1.14): PSID one-year-measure correlation 0.20 (elasticity ~0.07) vs correlations OVER 0.5 with longer-run income measures - the transitory-income attenuation Solon 1992 corrected; our multi-year IGE band [0.40,0.60] sits inside their long-run range and the one-year number documents why naive single-year IGEs bias low
grandchild earnings → greatgrandchild earnings Full identifier `grandchild_earnings->greatgrandchild_earnings`	0.55 [0.4, 0.6]	gap_multiplier	structural	lindahl2015;adermon2018 Sweden 3-generation admin same IGE band; weakest-identified layer - Lindahl 2015 shows persistence beyond the middle generation survives conditioning; BEHRMAN-TAUBMAN 1990 CROSS-CHECK (v1.14): PSID one-year-measure correlation 0.20 (elasticity ~0.07) vs correlations OVER 0.5 with longer-run income measures - the transitory-income attenuation Solon 1992 corrected; our multi-year IGE band [0.40,0.60] sits inside their long-run range and the one-year number documents why naive single-year IGEs bias low
ipv exposure → daughter violence odds Full identifier `ipv_exposure->daughter_violence_odds`	2.5 [2, 3]	odds_ratio	boundary	widom1989;ehrensaft2003 US prospective cohorts cycle-of-violence transmission; upstream household_ipv incidence link queued (aizer2010)
displacement → local service jobs Full identifier `displacement->local_service_jobs`	5 [1.6, 5]	level_ratio	conditional	moretti2010 US metros LEVEL ratio: ~5 local service jobs per high-tech job lost (1.6 for manufacturing); applied as count*n, never chained as a multiplier
school spending → child earnings Full identifier `school_spending->child_earnings`	1.07 [1.03, 1.1]	gap_multiplier	boundary	jackson2016 US school-finance reforms per +10% per-pupil sustained 12y -> ~+7% adult earnings; module normalizes by exposure fraction
youth wages → youth crime Full identifier `youth_wages->youth_crime`	-1 [-1.5, -0.5]	percent_delta	boundary	gould2002;grogger1998 US counties / young men elasticity of youth crime participation w.r.t. local wages (~-1); band = rounding band, exact CI extraction queued
wage ratio → household ipv Full identifier `wage_ratio->household_ipv`	-0.813 [-1.45, -0.18]	log_elasticity	boundary	aizer2010 US counties 1990-2003, women's assault hospitalizations elasticity of ln(IPV) w.r.t. female/male wage ratio; AER Table 2 col 3 (lagged-DV spec), 95% CI from clustered SE 0.317. SIGN: male displacement RAISES the ratio and so LOWERS IPV against women (Aizer bargaining); wiring this into the daughter chain needs incidence weights for who is displaced - deliberately not composed here
displacement → infant birth weight Full identifier `displacement->infant_birth_weight`	0.9541 [0.912, 0.9981]	gap_multiplier	conditional	lindo2011 US PSID children 1985+, husband displacement, mother fixed effects Table 2 col 3 log birth weight -0.047 (SE 0.023) -> level multiplier exp(); 95% CI band; robustness cols 2-5: -0.041 to -0.047; effects concentrated below median birth weight (Fig 2); read via IZA DP 5213
import shock → non displaced wage spillover Full identifier `import_shock->non_displaced_wage_spillover`	-0.822 [-1.3042, -0.3398]	percent_delta	boundary	adh2013 US commuting zones 1990-2007, noncollege workers OUTSIDE manufacturing, 2SLS state-clustered SE REPLACED-queue-24 LANDED v1.12: THE SPILLOVER LINK - wage response of NON-displaced workers per $1,000/worker decadal import exposure. Table 7 Panel B col 6: -0.822 (SE 0.246) log pts per $1k -> 95% CI +/-1.96SE. All-workers nonmfg col 4: -0.761 (0.260). Pooled male TOTAL CZ wage response incl. the directly displaced (Table 6 col 2): -0.892 (0.294) - use that for the V1 total-response cross-check (V1 earnings row undershot ~5x; this link closes the gap). Authors caveat (note 39): CZ-average wages may understate composition-constant losses. Boundary-applied coefficient: NEVER chained
family income shock → child achievement sd Full identifier `family_income_shock->child_achievement_sd`	0.061 [0.0157, 0.1063]	sd_delta	conditional	dahl2012 US children (NLSY79 mothers) ages 5-14, EITC instrumental variables queue-8 LANDED v1.12: +$1,000 (year-2000 $) current family income -> +0.0610 SD combined math-reading achievement. AER Table 3 col (i), SE 0.0231, N=8609 published (8608 in NBER w14599; verified against published AER tables 2026-09-07, coefficient+SE identical), two-year differences (child FE); 95% CI = point +/- 1.96 SE. Sign: an income LOSS -> achievement LOSS (apply against a negative shock). Robustness Table 4 specs A-F: 0.053-0.086 all sig; lasting-effects Table 5: small/imprecise (effect mostly contemporaneous). PARALLEL stream: boundary-applied, never chained with the earnings-gap multiplier; CARNEIRO-LOPEZ-GARCIA-SALVANES-TOMINEY 2021 TIMING CORROBORATION (v1.15, Norway admin, fixed permanent income): income in EARLY (0-5) and LATE (12-17) childhood has high marginal productivity for child human capital; MIDDLE (6-11) is low; balanced profiles beat imbalanced. Structural-model quantities - deliberately NOT a composed row; it bounds which childhood window the model's income streams may claim to harm; bingley2026 CROSS-CHECK (v1.40): Danish plant-closure children (JEEA 2026) - displacement's math-achievement hit concentrates in below-median-income families and in breadwinner displacements; authors' mediation slope +0.0008 SD math per +1,000 DKK/year family income (p<.01); timing concentrated below age 5, matching the early/late-window productivity finding recorded above
family income shock → life expectancy years Full identifier `family_income_shock->life_expectancy_years`	0.1333 [0.1167, 0.15]	life_years	boundary	chetty2016 US adults age 40, household earnings P15-P20 ($14k-$20k 2012 dollars), men+women averaged queue-19 LANDED v1.12: LOCAL income-longevity slope for the earnings->life-years conversion. DERIVED FROM the paper own concavity example (Results): $14k->$20k (P15 vs P20), $161k->$224k (P90/P95), $224k->$1.95M (P95/P100) each carry +0.7 to 0.9 years; taking the P15-P20 bracket (the displaced-worker income range): 0.8/6 = 0.1333 y per $1k, band 0.7/6 - 0.9/6. The gradient is approximately LINEAR in income PERCENTILE and CONCAVE in dollars - this slope applies near the working-class income range and OVERSTATES per-$ gains higher up. HONESTY: association, not causal - authors state unmeasured confounding means causal effects are likely SMALLER; the model causal deaths stream stays anchored on sullivan2009 rate ratios, this row is a conversion/summary factor only. Sex-specific anchors from the same text: P1-P99 gap 14.6y men [14.4,14.8], 10.1y women [9.9,10.3]. Boundary-applied: NEVER chained
displacement → college enrollment Full identifier `displacement->college_enrollment`	0.9894 [0.9848, 0.9939]	rate_ratio	conditional	hilger2016 US children of ~7M laid-off fathers, ages 18-22 enrollment, 1999-2009 administrative data queue-11 LANDED v1.13: paternal layoff -> college enrollment -0.432pp (SE 0.094) on a 40.66% base = -1.06% relative; multiplier = 1-0.432/40.66, 95% CI from 1.96SE. THE HONESTY LOADSTONE: single-difference closure estimates are 13x larger (-5.54pp, Table 4 col 1) but Hilger DD on closures is wrong-signed (+0.00569, SE 0.0032) - firm-closure designs carry assortative-selection bias, and his footnote 31 names oreopoulos2008 and sullivan2009 as exposed. Our child-earnings and mortality anchors ARE closure/displacement designs: tension recorded, V2 cross-check queued, child-earnings band NOT silently changed
displacement → parental income shortrun Full identifier `displacement->parental_income_shortrun`	0.864 [0.8568, 0.8712]	gap_multiplier	conditional	hilger2016 US laid-off fathers, annual income in the layoff event window (base $59,832) queue-11 companion LANDED v1.13: layoff -> parental income -$8,138.09/yr (SE 219.53) on base $59,832 = -13.60%; multiplier + CI from 1.96SE. SHORT-RUN event-window income - do not compose into worker_earnings (long-run JLS band); useful to scale exposure magnitudes in US data. STEVENS-SCHALLER 2011 CROSS-CHECK (v1.20): US SIPP family income -10.9% (SE 2.8pp) one-or-more years after the head's involuntary job loss (Table 3, read via NBER w15480) - corroborates this row's short-run income-loss magnitude in a second US sample; row values unchanged
unemployment rate → property crime Full identifier `unemployment_rate->property_crime`	5.018 [2.795, 7.241]	percent_delta	boundary	raphael2001 US states 1971-97, N=1,293 state-years, 2SLS defense-contracts + oil-cost instruments queue-10 LANDED v1.13: +1pp state unemployment -> +5.018% property crime (SE 1.134), Table 5 "All property crime" linear+quadratic-trends 2SLS (95% CI = 1.96SE); overidentification PASSES for property crime in all 2SLS specs. OLS full-spec 1.635 (0.182); the paper summary quotes OLS 1.6-2.4% per pp. Violent crime: null/mixed (2SLS insig 2 of 3 specs, overID rejected in the third) - deliberately NOT landed. Lin 2008 second anchor still queued for the band. Boundary-applied: NEVER chained
eitc exposure → adult earnings early Full identifier `eitc_exposure->adult_earnings_early`	564 [83.996, 1044]	usd	boundary	bastian2018 US PSID children, EITC exposure ages 13-18, PSID 1968-2013, state-clustered SE queue-bonus LANDED v1.14: +$1,000 EITC exposure during ages 13-18 -> +$564 annual earnings at ages 22-27 (SE 244.9, p<0.05; Table 2 col 6, 95% CI = 1.96SE; 2013 dollars). First stage: $1k exposure ~ $12,000 family income over the window (~$2k/yr x 6y) -> implies ~+$47 per $1k income. IV-scaled version (Table 6 col 6: +$57.2, SE 32.4) is significant only at 10% - recorded here, NOT landed as its own row to avoid implying precision it lacks. TIMING: exposure ages 0-12 ~ null on education outcomes (F-test p=0.006 for HS: teen exposure dominates) - corroborates Carneiro et al. timing; the adolescent income stream is the one that bites. Education outcomes (HS grad +1.2pp on 0.92, college grad +1.3pp on 0.31) recorded, not landed. Sign note: displacement INCOME LOSS applies this row negatively. Boundary-applied: NEVER chained; CARNEIRO-LOPEZ-GARCIA-SALVANES-TOMINEY 2021 TIMING CORROBORATION (v1.15, Norway admin, fixed permanent income): income in EARLY (0-5) and LATE (12-17) childhood has high marginal productivity for child human capital; MIDDLE (6-11) is low; balanced profiles beat imbalanced. Structural-model quantities - deliberately NOT a composed row; it bounds which childhood window the model's income streams may claim to harm
male earnings → marital fertility Full identifier `male_earnings->marital_fertility`	1.24 [0.397, 2.083]	log_elasticity	boundary	kearney2018fracking US PUMAs, fracking boom 1997-2012, women 18-34, IV (shale reserves), PUMA-clustered SE queue-7 LANDED v1.14: elasticity of ln(marital birth rate) w.r.t. ln(male earnings per capita) = 1.24 (SE 0.43), fracking boom IV (Table 8 Panel B col 2, women 18-34, N=32,704 PUMA-years, 1997-2012), 95% CI = 1.96SE. DECLARED CAVEATS: first-stage F=11.8 (weak-ish instrument); nonmarital births carry the SAME elasticity (col 3) - income moves births, NOT marriage (Table 4 finds no marriage response); cross-context anchor: Appalachian coal boom marital elasticity 0.75 (0.23) - the band brackets both. Read via NBER w23408; REStat 100(4):678-690 is the version of record. MODEL DIRECTION: this is a POSITIVE-income-on-births estimate - a displacement earnings LOSS applies it negatively (fertility falls). Boundary/elasticity composition; never a level chain. THORNTON 1980 CROSS-CHECK (v1.18): cross-generation fertility transmission, PSID two-generation OLS (Population and Environment 3(1):51-72, Table 3) - ACTUAL family size transmits near-null (siblings-of-husband -> total expected fertility: zero-order .070, standardized .058 with education controls, ns; parity sign-inconsistent across 1972 (-.044) and 1974 (.032)); IDEAL family size (preferences) transmits strongly (.282* zero-order, .237* with controls; unstandardized: +1 child in parental ideal -> child ideal +0.15, expected +0.08); parental economic status correlates negatively with child parity, attenuated by education controls. OBSERVATIONAL correlations - no parameter row licensable (CITING 1: quasi-experimental designs only); the model carries NO cross-generation fertility multiplier and this note is the recorded reason. Sample: couples married by 1971-1973 (N 293-566, Table 1)
import shock → gop win probability Full identifier `import_shock->gop_win_probability`	24.08 [0.423, 47.737]	percent_delta	boundary	autor2020polarization US county-congressional-district cells 2002-2010, N=3,772, 2SLS full controls queue-16 LANDED v1.15: change in the probability of electing a Republican to the House, in percentage points, per $1k/worker decadal import exposure = 24.08 (SE 12.07; t=2.00), Table 4 col 6, 2002-2010, N=3,772 county-district cells, 2SLS full controls (spec = Table 3 col 5), 95% CI = 1.96SE. Authors' own scaling: x 0.49 (p75-p25 exposure gap) = 11.8pp. HONEST SCOPE: the two-party VOTE-SHARE columns are NULL (all districts -1.08, SE 5.98; competitive +6.10, SE 4.93; solid-GOP -6.24, SE 3.93 - re-sorting, not uniform right shift); the win-probability effect emerges only from the 2010 mid-term (null 2002-2008, Feigenbaum-Hall consistency). Boundary-applied: NEVER chained
male job loss rate → child maltreatment Full identifier `male_job_loss_rate->child_maltreatment`	6 [3.65, 8.35]	percent_delta	boundary	lindo2018maltreatment California counties, child-maltreatment reports, county-year panel, county-clustered SE queue-14 LANDED v1.16: +1pp male mass-layoff rate -> +6.0% child-maltreatment reports (SE 1.2), Table 3 Panel A col 3 (preferred: county controls + county linear trends), Poisson semi-elasticity, California counties, N=817 county-years, county-clustered SE, 95% CI = 1.96SE. GENDER SCOPE (declared): the FEMALE mass-layoff coefficient is opposite-signed (-6.5% in the same spec) per the authors' family-time-use framework - this row models the male-shock channel, matching the displacement pool's male skew; do NOT apply to female-shock scenarios. Overall unemployment rate is NEGATIVELY associated (-2.1 to -2.4%, Table 2) - endogeneity warning the authors themselves flag; the mass-layoff quasi-experiment is the identification. Cross-stream corroboration: raphael2001 (male job loss -> property crime). Read via NBER w18994; JPubE 163:77-98 is the version of record. Boundary-applied: NEVER chained
displacement → grade retention hazard Full identifier `displacement->grade_retention_hazard`	1.1473 [1.0119, 1.2827]	rate_ratio	conditional	stevens2011 US children 5-19 (SIPP 1996/2001/2004 panels), head-of-household involuntary job loss (fired/discharged, employer sold/bankrupt, slack work), child fixed effects, N=167,449 child-years / 2,170 displaced children / 1,196 job losses queue-12 LANDED v1.20. QUEUE TARGET CORRECTED: the paper has NO math/reading test scores - the outcome is grade retention, the authors' own proxy for academic difficulties (correlated with test scores per McCoy-Reynolds 1994; causal-retention literature Jacob-Lefgren 2009 cited). VENUE CORRECTED: version of record is Economics of Education Review 30(2):289-299 (Crossref-verified); the queue and prior bib said Sociology of Education 84(3):201-223 - no such Crossref record exists. Full text read via NBER w15480. Anchor: Table 4 col 1 (full sample, child FE), head's job loss one or more years ago +0.0081 (SE 0.0038) on the 0.055 sample average = +0.8pp ~ +15% relative; multiplier = 1 + 0.0081/0.055 = 1.147, 95% CI from 1.96SE scaled by the same base. TIMING STRUCTURE (the identification): current-year coefficient 0.0032 (0.0040) is null - children are not coded as repeating until the second attempt year, so only the lagged coefficient carries the causal effect; no pre-trend. HETEROGENEITY (Table 5): effects concentrated where the head has HS education or less (+0.0099, 0.0058) vs some-college +0.0026 (0.0039, ns) - the model's displaced population skews less-educated, so the full-sample point is if anything conservative for them. Income mechanism measured in-sample: family income -10.9% (SE 2.8pp) 1+ years after loss (Table 3). PARALLEL STREAM: grade retention is a school-progress indicator in hazard space, NOT child_achievement_sd - never compose with the dahl2012 income mechanism row or the earnings-gap multiplier. Boundary-applied: NEVER chained
eviction order → emergency shelter use Full identifier `eviction_order->emergency_shelter_use`	4.778 [1.076, 8.48]	rate_ratio	boundary	collinson2024 US eviction-court cases (compliers), Cook County + NYC, heads 18-55, judge-leniency IV (first-stage F=934/288) queue-17 LANDED v1.21. QUEUE TARGET CORRECTED: version of record is collinson2024, QJE 139(1):57-120 (Crossref-verified) - the queue named the Collinson & Reed 2018 WP, which the QJE paper subsumes; read via NBER w30382 (rev. July 2023). Emergency-shelter use in the FIRST YEAR after filing: IV +3.4pp (SE 1.7) on the 0.9% non-evicted base = rate ratio (0.009+0.034)/0.009 = 4.78; 95% band = 1.96SE on the delta rescaled by the same base. Table V col 3 (combined equal-weighted two-city estimates; Cook County +2.3pp / NYC +4.6pp separately - NYC larger, consistent with right-to-shelter). TIMING: the acute spike is year-1 concentrated - year-2 emergency-shelter IV is null (-0.1pp, SE 1.3) but ANY homelessness services stays elevated in year 2 (+3.6pp, SE 1.5, on the 1.2% base) - transition out of acute shelter into persistent service contact. CROSS-CHECK: the superseded NYC-only C&R 2018 design measured shelter APPLICATION (right-to-shelter, broader): IV +12.7pp (SE 6.2) on a 3.4% application base - different outcome, same direction. PARALLEL STREAM: hazard-space outcome; NEVER composed with the worker_earnings multiplier or chained into the child streams. Boundary-applied: NEVER chained
eviction order → eviction earnings response Full identifier `eviction_order->eviction_earnings_response`	0.8574 [0.7444, 0.9705]	gap_multiplier	boundary	collinson2024 same design as the shelter row queue-17 LANDED v1.21: evicted tenants' average QUARTERLY earnings, quarters 5-8 post-filing: IV -$613 (SE 248) on the $4,300 non-evicted mean = -14.3%; 95% CI = 1.96SE; stored as a gap multiplier: 1 - 613/4300 = 0.8574 (band 1 - CI/4300), 2016 USD in the derivation above. Year-1 window (quarters 1-4) is -$323 (SE 175), NOT significant - recorded, not landed (the stevens2011 timing pattern: the sustained window carries the identified effect). OLS -$269*** is SMALLER than IV - the authors read complier effects as larger than the average evicted tenant's. SUBGROUPS (Table VI): concentration among female (-$767, SE 295) and Black (-$931, SE 307) tenants in year 2; formal equality tests do not reject, so the full-sample point stands with the concentration noted. REFUSAL WIRING: the desmond2016 matched +11-22pp JOB-LOSS estimate is NOT encoded (CITING §1 class + collinson2024 refutes the size: employment IV -1.5pp (SE 2.1) year 1 / -1.8pp (SE 2.7) year 2, both null); see SPEC §10. PARALLEL STREAM: quarterly-earnings LEVEL response for the eviction-court population - NEVER composed into the worker_earnings displacement multiplier. Boundary-applied: NEVER chained
foreclosure order → house price gap Full identifier `foreclosure_order->house_price_gap`	0.99 [0.9875, 0.9925]	gap_multiplier	boundary	campbell2011 US all-type house transactions, Massachusetts 1987-2008 (1.78M sales; Warren Group), reference distance 0.05 miles queue-18 LANDED v1.22: forced-sale price spillover. POINT is the authors' PREFERRED estimate: each foreclosure ~0.05 miles away lowers the price of a house by about 1% (lead-vs-lag difference-in-difference in the Leigh-Rockoff style: foreclosures within 0.1mi before the sale vs within 0.1mi after, inner-circle 0.1mi vs outer-ring 0.25mi; zero distance -2%, 0.05mi -1%). Stored as a gap multiplier: 1 - 0.01 = 0.99. BAND: the preferred DiD reports NO standard error, so the band is the rounding band (point +/-25%, CITING section 3 rule 3 phrasing) with the SE-carrying hedonic associates recorded beside, NOT averaged in: per foreclosure within 0.25mi year-before-sale -1.8% (col 1), -1.1% controlling recent neighborhood unforced prices (col 2); 0.1mi distance-weighted at zero distance -9.1% (col 1) / -7.3% (col 2) - clustering-inflated per the authors' own diagnosis, recorded to bound the same-location tail (their 99.9th-percentile dummies reach -21% to -26%). PERSISTENCE (Table 14): lagged foreclosures keep comparable coefficients a year out - spillovers are not transitory liquidity discounts. HETEROGENEITY (Table 13): larger for single-family and condos, absent for multifamily; larger for cheap houses in cheap zipcodes (vandalism mechanism). AGGREGATE (their 2008Q1 calculation): typical foreclosure cost $46,000 on the foreclosed house + $406,000 total on neighbors (col-2 model) or $110,000 (DiD). IDENTIFICATION LIMIT (authors' own statement): no instrument found; estimates are NOT structural - timing (lead vs lag) and distance gradient are the only quasi-experimental handles, so this lands as a PARALLEL-STREAM boundary-applied level response, never composed into the earnings/child chains (housing-contagion composition is a queued P3 item). CLASS/VERSION CAVEAT: read via NBER w14866 (April 2009 pre-publication draft); the published AER 2011 abstract confirms the preferred estimate (about 1% at 0.05mi) and the 27% foreclosure discount - table-level numbers are the WP's, tier therefore EXACT-results not EXACT. QUEUE TARGET CORRECTED: venue is AER 101(5):2108-31, NOT QJE 126; Immergluck & Smith 2006 (Housing Studies) is their CRIME paper - the price-spillover study is Immergluck & Smith 2005 Woodstock report / Housing Policy Debate, unread, recorded as a cross-check lead only
household ipv → child internalizing sd Full identifier `household_ipv->child_internalizing_sd`	0.48 [0.39, 0.57]	sd_delta	boundary	evans2008 children/adolescents 18 or younger exposed to interparental physical violence (second-generation studies, 1990-2006); k=58 samples, N=7,602 queue-21 LANDED v1.23: meta-analytic internalizing effect. Random-effects model, k=58 samples (60 studies / 61 samples; second-generation 1990-2006 literature), N=7,602; d = .48 (SE .04), 95% CI .39-.57, z = 11.25; homogeneity ns (Q(56) = 70.61, p > .05) - so no moderator split landed. CROSS-CHECKS recorded, never averaged (CITING section 5): Kitzmann et al. 2003 d = -.50 (group design, published+unpublished 1978-2000); Wolfe et al. 2003 d = .38 (zr = .19 converted). Gender ns (boys .44 vs girls .39, QB = .34, p = .56). PARALLEL STREAM: boundary-applied exposure contrast - NEVER composed into the child_earnings/IGE chains or the achievement SD row; gender/age nulls declared. CLASS: meta-analysis sets points per CITING section 1 (band = reported CI); exposure mostly dichotomous and studies mostly cross-sectional - the ordering caveat is the authors' own. TRAUMA REFUSAL: d = 1.54 (SE .59, CI .38-2.71) NOT landed - k=6, heterogeneous (Q(5) = 18.35, p < .01), the authors themselves say interpret with caution. Read via the accepted manuscript (Nebraska DigitalCommons psychfacpub/321, Elsevier permission copy via Wayback mirror of the WAF-blocked PDF)
household ipv → child externalizing sd Full identifier `household_ipv->child_externalizing_sd`	0.47 [0.38, 0.56]	sd_delta	boundary	evans2008 children/adolescents 18 or younger exposed to interparental physical violence (second-generation studies, 1990-2006); k=53 samples, N=7,200; pooled across gender - boys .46 / girls .23 queue-21 LANDED v1.23: meta-analytic externalizing effect. Random-effects model, k=53 samples, N=7,200; d = .47 (SE .05), 95% CI .38-.56, z = 10.11; HETEROGENEOUS (Q(52) = 80.35, p < .01) - moderators examined. CROSS-CHECKS recorded, never averaged (CITING section 5): Kitzmann et al. 2003 d = -.50 (group design, published+unpublished 1978-2000); Wolfe et al. 2003 d = .38 (zr = .19 converted). GENDER MODERATOR (significant, scope declared): boys d = .46 vs girls d = .23 (QB = 4.11, p < .05) - the pooled .47 pools a strong boy effect and a weak girl effect; age and age-by-gender ns (adolescent cells underpowered, k=4). PARALLEL STREAM: boundary-applied exposure contrast - NEVER composed into the child_earnings/IGE chains or the achievement SD row; gender/age nulls declared. CLASS: meta-analysis sets points per CITING section 1 (band = reported CI); exposure mostly dichotomous and studies mostly cross-sectional - the ordering caveat is the authors' own. TRAUMA REFUSAL: d = 1.54 (SE .59, CI .38-2.71) NOT landed - k=6, heterogeneous (Q(5) = 18.35, p < .01), the authors themselves say interpret with caution. Read via the accepted manuscript (Nebraska DigitalCommons psychfacpub/321, Elsevier permission copy via Wayback mirror of the WAF-blocked PDF)
local unemp shock → household ipv Full identifier `local_unemp_shock->household_ipv`	1.58 [1.49, 1.58]	rate_ratio	boundary	schneider2016 mothers of the Fragile Families birth cohort (1998-2000 births in 20 large US cities), Waves 3-5 pooled person-waves (2001-2010), all-mothers sample including unpartnered; N=10,656 area models / 10,584 distress models; UR shock 1.0=flat, 2.0=doubling queue-22 LANDED v1.24: Table 3, violent-or-controlling outcome. Logistic coeff .454* (p<.05) per unit of the 12-month UR change (1.0 = doubling); LDV .401, FE .406 - the band is the exp() range across the three specs (1.49-1.58); the point is Model 1 (logit), the paper's headline spec. Authors' own predictions: UR worsening 50% -> prevalence 10%->12%; UR doubling -> 14% (prevalence ratios 1.2/1.4, recorded here). Controlling-only coefficients: .472/.482/.453* (logit/LDV/FE). IDENTIFICATION: area-level unemployment is exogenous to individual traits; LDV and individual-FE specs; cluster-robust SEs by city. LEVEL-NULL discipline: the UR LEVEL (12-month average, Table 8) is null on all three outcomes - only the 12-month CHANGE moves abuse; recorded so no surface wires the level. Violent-only is null for BOTH unemployment measures; hardship->violent is the authors' own reverse-causality concern (violent behavior dis-orders household finances). FE: couple-unemployment magnitude unchanged, hardship magnitude reduced (controlling) but still significant. CLASS: observational regression coefficients = points (CITING section 1); no SEs printed in the text, bands declared. PARALLEL STREAM: boundary-applied shock responses - NEVER chained into the earnings/IGE chains
household hardship → household ipv Full identifier `household_hardship->household_ipv`	2.14 [1.61, 2.68]	rate_ratio	boundary	schneider2016 mothers of the Fragile Families birth cohort (1998-2000 births in 20 large US cities), Waves 3-5 pooled person-waves (2001-2010), all-mothers sample including unpartnered; N=10,656 area models / 10,584 distress models; hardship = 1+ of 10 material-hardship items, past 12 months queue-22 LANDED v1.24: Table 1 Model 1 prose predictions - hardship -> violent-or-controlling prevalence 15% vs 7% (point 2.14); controlling-only 13% vs 7%; violent-only 2% vs 0.05% (reverse-causality-flagged above). Table 1 coefficients (scrambled by text extraction; ~OR 2.15-2.31 across logit/LDV/FE) NOT used for the band - the band is the declared rounding band +/-25% (campbell2011 precedent). IDENTIFICATION: area-level unemployment is exogenous to individual traits; LDV and individual-FE specs; cluster-robust SEs by city. LEVEL-NULL discipline: the UR LEVEL (12-month average, Table 8) is null on all three outcomes - only the 12-month CHANGE moves abuse; recorded so no surface wires the level. Violent-only is null for BOTH unemployment measures; hardship->violent is the authors' own reverse-causality concern (violent behavior dis-orders household finances). FE: couple-unemployment magnitude unchanged, hardship magnitude reduced (controlling) but still significant. CLASS: observational regression coefficients = points (CITING section 1); no SEs printed in the text, bands declared. PARALLEL STREAM: boundary-applied shock responses - NEVER chained into the earnings/IGE chains
couple unemployment → household ipv Full identifier `couple_unemployment->household_ipv`	1.3 [1.04, 1.63]	rate_ratio	boundary	schneider2016 mothers of the Fragile Families birth cohort (1998-2000 births in 20 large US cities), Waves 3-5 pooled person-waves (2001-2010), all-mothers sample including unpartnered; N=10,656 area models / 10,584 distress models; either partner unemployed and looking queue-22 LANDED v1.24: Table 1 Model 1 prose predictions - couple unemployment -> violent-or-controlling prevalence 13% vs 10% (point 1.3); controlling-only 12% vs 9%; violent-only ns. Table 1 logit coeff .404** (OR ~1.50); FE magnitude unchanged (stronger survival than hardship). Disaggregation (results on request): one-partner and both-partner unemployment both significant in logit/LDV, n.s. in FE; male-partner coefficients ~50% larger than mother-own for controlling outcomes (Table 2) but male/female-specific FE estimates n.s. IDENTIFICATION: area-level unemployment is exogenous to individual traits; LDV and individual-FE specs; cluster-robust SEs by city. LEVEL-NULL discipline: the UR LEVEL (12-month average, Table 8) is null on all three outcomes - only the 12-month CHANGE moves abuse; recorded so no surface wires the level. Violent-only is null for BOTH unemployment measures; hardship->violent is the authors' own reverse-causality concern (violent behavior dis-orders household finances). FE: couple-unemployment magnitude unchanged, hardship magnitude reduced (controlling) but still significant. CLASS: observational regression coefficients = points (CITING section 1); no SEs printed in the text, bands declared. PARALLEL STREAM: boundary-applied shock responses - NEVER chained into the earnings/IGE chains
unemployment status → mental health sd Full identifier `unemployment_status->mental_health_sd`	0.51 [0.47, 0.54]	sd_delta	boundary	paul2009 involuntary unemployed vs employed adults (26 countries, 1950-2004); cross-sectional meta k=315 samples, N=209,379, outliers excluded queue-20 LANDED v1.25: overall cross-sectional random-effects meta, outliers-excluded row (k=315 samples, N=209,379; 237 cross-sectional + 87 longitudinal studies, 26 mostly-Western countries, 1950-2004). Band = reported CI (CITING section 1 meta class; random-effects, conservative model by the authors' own choice). ALL-STUDIES row beside: d = 0.54 (SE .0184, CI .50-.57, k=323, N=458,820); outlier exclusion reduced it 0.54->0.51. HETEROGENEOUS (H = 1.91-2.70 across indicators; Q p<.001) - moderators examined. CASE-RATE translation (practical-importance anchor): psychological-problem case rates 34% (CI .31-.38, k=76) among unemployed vs 16% (CI .14-.18, k=74) among employed. CAUSALITY ANCHORS recorded beside (never averaged - CITING section 5): factory-closure natural experiments d = .38 (CI .25-.51, k=27, N=4,246, Qb=3.99 p<.05 vs other designs .55 - the causal-clean subset); longitudinal job loss d = .19 (CI .10-.29, k=19), reemployment d = .35 (CI .42-.28, k=45), RETEST-CORRECTED 0.25/0.29 (the repeated-measurement artifact explanation for the earlier metas' puzzling asymmetry); school->unemployment d = .10 ns. SELECTION effects (reverse causation) small but significant: distress->job loss d = .23, distress->worse reemployment d = .15, school .08 - declared: selection runs the OTHER way and is weak, not the driver. MODERATORS (scope-declared): men stronger (beta .22/.20 controlled), blue-collar stronger (beta .19 controlled; partially confounded with gender, reduced to .13 marginal), duration linear beta .13/.15 + curvilinear - distress peaks at 9 months (d = .73), stabilizes elevated in year 2, renewed worsening after 29 months UNSTABLE (k=5, cubic marginal in controlled analysis - not landed); country moderators: low-GDP d = .62 vs .49 (Qb=11.84), high-Gini .57 vs .48 (Qb=4.97), weak protection .58 vs .46 (Qb=8.61, confounded with Gini per the authors); age U-shape UNSTABLE (vanished when marital status controlled - not landed); marital status ns. PARALLEL STREAM: boundary-applied status contrast - NEVER composed into the child_earnings/IGE chains or the achievement SD rows. Read via owner-downloaded ScienceDirect PDF; marcus2013 CROSS-CHECK (v1.38): SOEP plant-closure entropy-balancing DiD, own mental health -2.72 MCS points = -0.272 SD (SE 0.74) ~11 months after closure - inside the causal-clean factory-closure subset (.25-.51), consistent; schallerstevens2015 CROSS-CHECK (v1.40): US MEPS (J Health Econ 43:190-203), preferred individual-FE + baseline-trend spec - P(fair/poor mental health) +1.39pp on 3.4% base (+40.9% relative), depression/anxiety +1.64pp on 7.3%; probability units, not SD-convertible - directional corroboration from US survey data; no chronic-condition incidence increase survives baseline-trend controls
import shock → radical right vote share Full identifier `import_shock->radical_right_vote_share`	13.2 [3.2, 23.2]	percent_delta	boundary	colantone2018 15 Western European countries, 76 legislative elections 1988-2007, 198 NUTS-2 regions; district-election panel N=7,782 (IV), country-year FE, SE clustered NUTS-2 x year queue-15 LANDED v1.26: IV Table 1 col 10, radical-right vote share. Coefficient .132* (SE .051) in vote-share FRACTIONS per 1.0 ($1k/worker change in Chinese import exposure over the past 2 years) = 13.2 pp per $1k/worker of exposure (SE 5.1), CI = 1.96 SE. AUTHORS' OWN SCALING: one SD of the 2-year shock (.133 = $133/worker) -> +1.7pp radical-right vote share, on an average RR share of 5% (SD 7%). IV (.132) > OLS (.041, SE .023): positive domestic demand shocks attenuate the OLS association (authors' interpretation). First stage: Chinese imports to the US as instrument, Kleibergen-Paap F = 19.2. Robustness (Table 3-4): historical-trajectory trend interactions (coefficients equal or larger), year-specific regional controls (larger), alternative exchange-rate IV, exports and FDI nulls, EU-import controls (EU15 negative on Nationalist Autarchy, EU12 positive on the radical right). NO-LEFT-WING discipline: protectionist left NULL (IV -.052, SE .047), protectionist left proper NULL (-.081, SE .042), liberal right NULL (-.017, SE .075) - NO polarization, which CONTRASTS with the Autor et al. 2020 US finding (declared on that row); pro-trade left NEGATIVE significant (-.134, SE .054) - the electorate abandons mainstream social-democratic parties; protectionist right POSITIVE .278*** (SE .094). INDIVIDUAL-LEVEL (ESS waves 1-4, Table 5-6): confirms the district evidence; NO significant group heterogeneity - interactions all n.s., the response is SOCIOTROPIC (service and public-sector workers respond too; only students sheltered, n.s. overall) - declared so no surface reads this row as a displaced-manufacturers-only effect, and so no surface quotes paul2009's person-level unemployment distress as the transmission mechanism here. WINDOW DISCIPLINE: the regression's shock is the 2-year window change; the row stores the same coefficient expressed per $1k/worker of cumulative exposure (13.2pp), which assumes each window's shock persists to the next election - the authors' 1-SD scaling reads directly off this coefficient. Boundary-applied: NEVER chained into the earnings/IGE chains. Read via the owner-downloaded PDF (the published AJPS doi, not the SSRN preprint doi the staging bib carried)
neighborhood exposure → child outcomes modifier Full identifier `neighborhood_exposure->child_outcomes_modifier`	0.037 [0.031, 0.043]	gap_multiplier	structural	chettyhendren2018 children of families moving across counties (US tax data, births 1980-88); county-level predictions of permanent residents' income ranks at age 24 queue-23 LANDED v1.30: annual childhood exposure effect gamma, Appendix Table V Column 1 (county level, baseline spec): gamma = 0.037 (SE 0.003), N = 595,244. Interpretation quoted verbatim from the paper: the impact of spending an ADDITIONAL YEAR of childhood in a county where children of permanent residents have 1 percentile point higher income ranks (at a given level of parental income). CZ-level baseline (Table II col 1) = 0.040 (SE 0.002, N = 1,553,021) recorded beside; county-level is the pin because places.csv is county geography and the authors note gamma is SMALLER at smaller geographies (selection stronger within CZs). DOSE: the engine multiplies by a declared 18-year complete childhood (DOSE_YEARS, place.py) = 0.666 pickup per 1 percentile gap vs national; the paper's own full-childhood extrapolation (birth movers to age 20 pick up ~80%, abstract + sec IV) brackets this. Dose years differ from the paper's age-23 window - DECLARED in place.py. Family FE = 0.033 (SE 0.011), time-varying controls 0.032 (SE 0.011), age-orthogonal 0.036 (SE 0.005): robust. Similar exposure effects on college attendance, marriage, teen birth (abstract) - only the income-rank effect is landed. Outcomes converge LINEARLY in exposure; place effects act through time spent, not age of arrival. Band = reported SE x 1.96 = [0.0312, 0.0428] rounded [0.031, 0.043]. APPLIED TO: child-window adult-outcome multipliers (child_earnings and the generation line), via the place multiplier; the displacement penalty and the place modifier compose MULTIPLICATIVELY - a declared modeling assumption, not a claim of the paper
unconditional income → child education years Full identifier `unconditional_income->child_education_years`	1.127 [0.247, 2.007]	edu_years	boundary	akee2010 children in Eastern Band of Cherokee area households previously in poverty (Great Smoky Mountains Study, age cohorts 9/11 at wave 1), treated by ~$4,000/yr per adult member for four pre/post childhood survey waves queue-25 LANDED v1.32: Table 4 col 1 (households previously in poverty), years of education at age 21 = +1.127 (SE 0.449, 1% level), OLS, N=1045 - the abstract's '+1 year at age 21 for the poorest households'. Per-treatment dose: four additional years of the ~$4k/y exogenous transfer; the dose lives in the exposure description, never a per-dollar chain step. SCOPE (never averaged away): full-sample coefficient is 0.379 (SE 0.447, n.s. - Table 3 col 1) because households NEVER previously in poverty show null/negative effects (Table 4 cols 3-4: -0.166 ns / +0.129 ns); the effect concentrates where poverty already binds. Robustness (fn18 predicted-poverty sample, N~600): 1.273 (1% level). GENDER OF RECIPIENT: mother-receipt drives it (Interaction 1 = 1.48, SE 0.606, 5% level; father-receipt n.s.) - consistent with Duflo 2003; declared. Companion HS-grad row: prob of HS graduation by age 19 +0.391 (SE 0.135, 1% level, T4 col 2) - the authors note the HS effect reads like a conditional-cash incentive (children become payment-eligible at 21 only via diploma-or-wait). MECHANISM (Table 8): parental supervision +0.062 (0.023)*/activities-with-mother +0.069 (0.024)*, maternal labor force participation NULL (Table 7: +0.069/−0.089 n.s.) - income, not employment, is the channel. OBSERVATIONAL CROSS-CHECK (barred from a point by CITING SS1) recorded on the queue-9 refusal: zg2012 Table S5 col 1 - prenatal-age-2 income +$10k/y among <$25k families -> ln adult annual earnings +0.63 (SE 0.21), age 3-5 +0.34 (0.21) n.s.
unconditional income → youth any crime Full identifier `unconditional_income->youth_any_crime`	-0.224 [-0.377, -0.071]	probability	boundary	akee2010 same population as the education row (previously-poor eligible households; arrests from administrative records merged at the individual level) queue-25 LANDED v1.32: Table 6 col 1, probit marginal effect on committing ANY crime (arrest) at ages 16-17 = -0.224 (SE 0.078, 1% level), N=1093 - the abstract's 'reduces having ever committed a minor crime by 22% at ages 16-17' (the abstract conflates: T6 shows the 22% on ANY arrest at 16-17 and 18% on ever-minor by 21; both landed as separate rows). AGE WINDOW (declared): the effect exists ONLY while the child is a MINOR - ages 18-19 (-0.068, SE 0.072) and 20-21 (+0.051, SE 0.075) are NULL, consistent with parental control ending at majority; no surface may generalize this row past 18. OFFENSE TYPE: the reduction is MINOR-CRIME ONLY (ever-moderate -0.002 n.s., ever-violent +0.002 n.s.) - an entry effect: income changes WHETHER a child enters criminality, not the NUMBER of crimes once entered (count regressions n.s.). Male-driven (little female criminality overall). Robust to a casino-distance interaction (fn21: -0.299 with distance interaction, 1% level). Boundary-applied, never chained
unconditional income → youth minor crime ever Full identifier `unconditional_income->youth_minor_crime_ever`	-0.179 [-0.353, -0.005]	probability	boundary	akee2010 same population as the education row (previously-poor eligible households) queue-25 LANDED v1.32: Table 6 col 4, prob of EVER committing a minor crime (disorderly conduct, trespassing, shoplifting) by age 21 = -0.179 (SE 0.089, 5% level), N=1045. Panel design: the treatment covers four pre/post childhood waves; effect = reduced ENTRY into criminality (counts-of-crimes regressions n.s. by category). Moderate/violent nulls beside (see the any-crime row); males-only analysis holds for minor crimes with diminished significance and strengthens for moderate. Band = 1.96 SE. Boundary-applied, never chained
youth crime conviction share → youth crime convicted Full identifier `youth_crime_conviction_share->youth_crime_convicted`	0.061 [0, 0.123]	probability	boundary	damm2014 male refugee children in Denmark assigned quasi-randomly to municipalities 1986-1998 (avg age at assignment 9; N=4,425) queue-24 LANDED v1.33 (queue target CORRECTED: the paper has NO clean exposure-duration elasticity - age at assignment is perfectly correlated with potential exposure years and the age-cohort splits confound duration with age, authors' own caveat; the landed number is the per-1pp conviction-SHARE exposure effect). Table 3 panel A, spec 5 (municipality FE - the authors' identification-focused spec isolating crime from fixed area characteristics): +0.043 probability points per 1 SD (0.70pp) of the area youth-crime conviction rate, SE 0.022 -> per 1pp: 0.061 (SE 0.031), band 1.96 SE; the band's lower edge rounds to 0.000 (the FE-spec SE leaves the interval at the no-effect line). Conservative spec 4 (full controls incl. police efficiency, no FE) recorded beside: 0.033/1pp (SE 0.017). RELATIVE scaling (paper prose): per 1pp -> +7-13% on the 46% male base conviction probability; count-of-convictions row beside: 0.169/0.70 = 0.241 convictions per 1pp (spec 5, SE 0.138). AGE-COHORT shape (declared): ages 10-14 at assignment most affected; <6 small/large SE. FEMALE NULL (declared, panel B - all specs n.s.). OPERATIVE-CHANNEL discipline (Table 6): the COMMITTED-crime rate is a NULL - only the share of CONVICTED YOUTH moves children; violent share is the operative crime type (property/drugs null, Table 5); co-national share channel (Table 7: co-nationals +0.023 per SD, immigrant shares null - social interaction via co-nationals). Boundary-applied, never chained. Read via Aarhus WP 2013-17 (Sept 2013 draft of the AER 104(6) version of record)
youth violent crime conviction share → youth crime convicted Full identifier `youth_violent_crime_conviction_share->youth_crime_convicted`	0.366 [0.143, 0.589]	probability	boundary	damm2014 same population as the overall-share row (male refugee children, quasi-random municipality assignment, N=4,425) queue-24 LANDED v1.33: Table 5 panel A col 2 (spec 5, municipality FE): +0.045 probability points per 1 SD (0.123pp) of the VIOLENT youth-crime conviction rate (mean 0.286%), SE 0.014 -> per 1pp: 0.366 (SE 0.114), band 1.96 SE. Spec 4 beside: 0.276/1pp (SE 0.089, 1% level). The violent share is the paper's OPERATIVE finding (abstract: 'in particular violent crimes'); property (0.029/0.528pp n.s.) and drug (null) shares recorded as nulls - the effect is peer CONVICTION exposure, not crime volume. FEMALE NULL (declared). Education null + 'active' decline by 23-25 recorded as longer-term context (Table 8), not landed. Boundary-applied, never chained; the violent share is a thin LEVEL (mean 0.286%) - never chained from a crime-count delta. Read via Aarhus WP 2013-17 (Sept 2013 draft of the AER 104(6) version of record)
displacement → circulatory mortality hazard Full identifier `displacement->circulatory_mortality_hazard`	1.54 [1.29, 1.83]	rate_ratio	conditional	browningheinesen2012 Danish full-time male workers age 20-60 with strong labor-market attachment displaced by plant closures 1986-2002 (base years 1985-2001; propensity-score weighted matched controls; Danish full registers 1980-2006) queue-sweep READ+LANDED v1.38: Browning & Heinesen 2012 (J Health Econ 31(4):599-616), Tables 6.1/6.2, cumulative hazard ratios years 1-4 after base year. Circulatory death. MI-or-stroke 1-4y 1.58 [1.24, 2.00] (Table 6.2); All-circulatory 1-20y 1.16 [1.07, 1.25] remains significant (Table 6.1); MI-or-stroke 1-20y 1.09 [0.97, 1.23] is null-compatible (Table 6.2). Year-1 peak beside. Propensity score weighting + non-parametric duration analysis; band = reported 95% CI on the cumulative hazard ratio. dist withheld: the reported hazard-ratio CI is asymmetric around the point; a symmetric sampler would fabricate the shape. Boundary-applied: NEVER chained. Source version rechecked 2026-10-14: 2011 AKF working paper hosted by VIVE, Tables 6.1/6.2 (PDF pages 23/25), SHA256 85465f4715493cce44be9fb78ddc1ecf08e838bc69b74858575e2fc6724ca164; the 2012 JHE citation identifies the published study, not a verified published-version extraction; REPLICATIONS (v1.43): bloemen2018 Dutch LPM circulatory +0.2453pp within 5y (SE 0.0975, p=.012) = +52.8% relative, cerebrovascular +0.0875pp (SE 0.0427) = +152.9% - the stress/circulatory channel corroborated across countries; eliasonstorrie2009 male <=4y circulatory 1.24 [0.90, 1.71] n.s. - direction consistent, imprecise; Publication tier corrected to canonical under CITING section 1: source is the 2011 working paper; published-version reconciliation remains queued.
displacement → alcohol mortality hazard Full identifier `displacement->alcohol_mortality_hazard`	1.62 [1.09, 2.41]	rate_ratio	conditional	browningheinesen2012 Danish full-time male workers age 20-60 with strong labor-market attachment displaced by plant closures 1986-2002 (base years 1985-2001; propensity-score weighted matched controls; Danish full registers 1980-2006) queue-sweep READ+LANDED v1.38: Browning & Heinesen 2012 (J Health Econ 31(4):599-616), Tables 6.1/6.2, cumulative hazard ratios years 1-4 after base year. Alcohol-related death (psychiatric + somatic: poisoning, addiction, liver, pancreas). The deaths-of-despair channel. 1-20y 1.15 [0.96, 1.38] n.s. - the effect concentrates in the first years. Year-1 peak beside. Propensity score weighting + non-parametric duration analysis; band = reported 95% CI on the cumulative hazard ratio. dist withheld: the reported hazard-ratio CI is asymmetric around the point; a symmetric sampler would fabricate the shape. Boundary-applied: NEVER chained. Source version rechecked 2026-10-14: 2011 AKF working paper hosted by VIVE, Tables 6.1/6.2 (PDF pages 23/25), SHA256 85465f4715493cce44be9fb78ddc1ecf08e838bc69b74858575e2fc6724ca164; the 2012 JHE citation identifies the published study, not a verified published-version extraction; Publication tier corrected to canonical under CITING section 1: source is the 2011 working paper; published-version reconciliation remains queued.
displacement → external cause mortality hazard Full identifier `displacement->external_cause_mortality_hazard`	1.53 [1.15, 2.04]	rate_ratio	conditional	browningheinesen2012 Danish full-time male workers age 20-60 with strong labor-market attachment displaced by plant closures 1986-2002 (base years 1985-2001; propensity-score weighted matched controls; Danish full registers 1980-2006) queue-sweep READ+LANDED v1.38: Browning & Heinesen 2012 (J Health Econ 31(4):599-616), Tables 6.1/6.2, cumulative hazard ratios years 1-4 after base year. External causes (incl. suicide, traffic). Suicide year-1 4.31 [1.64, 11.37] (15 deaths - wide), suicide-or-traffic 1-4y 1.74 [1.21, 2.48] (Table 6.2); 1-20y 1.20 [1.01, 1.42]. Cancer intervals include 1 across reported horizons (year-1 point 1.38, CI spans 1); null-compatible does not establish zero - cause-specific discipline: not all causes move. Year-1 peak beside. Propensity score weighting + non-parametric duration analysis; band = reported 95% CI on the cumulative hazard ratio. dist withheld: the reported hazard-ratio CI is asymmetric around the point; a symmetric sampler would fabricate the shape. Boundary-applied: NEVER chained. Source version rechecked 2026-10-14: 2011 AKF working paper hosted by VIVE, Tables 6.1/6.2 (PDF pages 23/25), SHA256 85465f4715493cce44be9fb78ddc1ecf08e838bc69b74858575e2fc6724ca164; the 2012 JHE citation identifies the published study, not a verified published-version extraction; REPLICATIONS + HETEROGENEITY (v1.43): eliasonstorrie2009 male <=4y external causes 2.07 [1.42, 3.02] and suicide 2.15 [1.28, 3.59] corroborate the deaths-of-despair channel; bloemen2018 DIVERGES - external-except-suicide -0.0389pp (SE 0.0176) and suicide +0.0063 n.s. - cause-specific patterns are country/period-heterogeneous, declared, never averaged; Publication tier corrected to canonical under CITING section 1: source is the 2011 working paper; published-version reconciliation remains queued.
displacement event → spouse mental health sd Full identifier `displacement_event->spouse_mental_health_sd`	-0.194 [-0.327, -0.061]	sd_delta	conditional	marcus2013 spouses of displaced workers, German couples (SOEP 2002-2010), 109 displaced-worker couples vs >14,000 entropy-balanced controls; MCS normed mean 50 SD 10 queue-sweep READ+LANDED v1.38: Marcus 2013 (J Health Econ 32(3):546-558), Table 3 spec 4 (entropy-balancing DiD, ~11 months after closure): spouse MCS -1.94 (SE 0.68) = -0.194 SD. OWN EFFECT beside: -2.72 (0.74) = -0.272 SD, recorded as a paul2009 cross-check (inside the causal-clean factory-closure subset); own-vs-spouse difference p=0.38 - the spillover is as large as the displaced worker's own effect. Husband-displaced spouses -2.09 (0.78), wife-displaced -1.67 (0.78). PLACEBO: closure WITHOUT unemployment own -0.77 n.s. - the unemployment experience, not the event, carries it (declared). Band = 1.96 SE. Read via SOEPpapers 488 (2012 WP draft of the JHE version of record; damm2014 read-via precedent). Boundary-applied: NEVER chained
displacement event → child achievement sd Full identifier `displacement_event->child_achievement_sd`	-0.0206 [-0.0402, -0.001]	sd_delta	conditional	bingley2026 children exposed to parental plant closure age 0-16 (average across exposure stages), Denmark, conditional on taking grade-9 exams; mathematics queue-sweep READ+LANDED v1.40: Bingley, Cappellari & Ovidi, JEEA (advance article 2026, DOI 10.1093/jeea/jvag048), Danish admin registers, displaced-parent plant closures matched to controls, children age 0-16 at closure. Point = Table 4 col 2 (preferred: baseline covariates, SE clustered at family) 'Treated X Exposure (age 0-16)': teacher grade in mathematics -0.0206 SD (SE 0.0100); band = 1.96 SE. BESIDE: exam test scores -0.0117 n.s. (SE 0.0101); Danish/language smaller and n.s. - the GPA decline is math-driven. TIMING STRUCTURE (the paper's headline; travels in notes - the model has no age-at-exposure selector): infancy (0-1) -0.0521 (SE 0.0154; 1.96SE -0.0823..-0.0219); early childhood (2-5) -0.0184 n.s.; mid (6-11) -0.0073 n.s.; late (12-16) -0.0274 (SE 0.0118) - concentrated before age 5 and again in adolescence. MECHANISM (declared, income channel): hit concentrates in below-median-income families and when the displaced parent is the breadwinner (main-earner infancy ~-0.07 both measures; maternal-breadwinner infancy -0.2 grades); authors' mediation: +1,000 DKK (~EUR 150) per year family income -> +0.0008 SD math (p<.01) - recorded as the dahl2012 cross-check; maternal time investment partially offsets. LEFT-TAIL beside: non-completion +0.54pp avg on 7.85% control mean (own row); upper-secondary enrolment at 17 -1.6pp infancy (6.5% of control mean). Read via IZA DP 16367 (Aug 2023 WP draft of the JEEA version of record; marcus2013/damm2014 read-via precedent). Conditional: transport assumes Danish institutions/welfare, plant-closure identification, grade-9 timing
displacement event → exam noncompletion hazard Full identifier `displacement_event->exam_noncompletion_hazard`	1.0688 [1.0014, 1.1362]	rate_ratio	conditional	bingley2026 same children/scope as the child_achievement_sd row; outcome = obtains NO grade-9 exam scores or teacher assessments (control mean 7.85%) queue-sweep LANDED v1.40, same read as the child_achievement_sd row. DERIVED RELATIVE (pp effect / control mean; stevens2011 derivation precedent): Table 2 col 2 (preferred) 'Treated X Exposure (age 0-16)' +0.0054 (SE 0.0027) on control mean 0.0785 -> 1 + 0.0054/0.0785 = 1.0688; band = 1.96 x SE/0.0785. AUTHORS' FRAME: explains 6-9% of observed non-completion (base ~7%; test-taking near-universal at 16). TIMING STRUCTURE, monotone by age at exposure: infancy (0-1) +0.99pp (SE 0.41) -> 1.126 [1.96SE 1.024..1.228]; early childhood (2-5) +0.76pp -> ~1.097; mid (6-11) +0.45pp n.s.; late (12-16) +0.30pp n.s.; exposure at age >=18 jointly zero (p=0.986 - identifying-assumption placebo). Completing exams BY AGE 16: infancy -1.44pp (SE 0.65). Concentrated in below-median-income families (no detectable effect above median); displaced father > mother, sons > daughters (qualitative). Conditional: Danish compulsory-school institution; left-tail outcome capturing the most vulnerable children
displacement event → hs completion Full identifier `displacement_event->hs_completion`	-0.037 [-0.0801, 0.0061]	probability	conditional	brand2014 children of single mothers, US NLSY79 (NLSY79 women's children), MATERNAL job displacement at child age 0-17 (PSM kernel TT); outcome = HS completion by age 19, control mean 0.760 LANDED v1.42 (user-directed node design 2026-09-13): Brand & Simon Thomas 2014 (AJS 119(4):955-1001), Table 3 kernel-matching TT: -0.037 (SE 0.022), p<.10 - the band honestly crosses zero. TIMING (Table 6 TT): age 0-5 -0.004 n.s.; 6-11 -0.016 n.s.; adolescence 12-17 -0.115** (SE 0.040; 1.96SE -0.1934..-0.0366) - an ADOLESCENCE effect; the always-exposure point stays the row because the model has no age-at-exposure selector. Beside: unmatched difference -0.053; TUT -0.052 - TUT>TT everywhere: effects LARGEST for children of LOW-displacement-propensity mothers (SM strata confirm; opposite gradient to bingley2026's income concentration). DESIGN DECLARED: propensity-score kernel matching = selection on observables, the weakest identification class in this parameter set (no quasi-experimental variation) - corroboration of the grade-retention/achievement streams, never an anchor. Read via PMC4372265 author manuscript of the published AJS version of record. Conditional: transport assumes the US single-mother NLSY79 cohort and the PSM design
displacement event → college attendance Full identifier `displacement_event->college_attendance`	-0.063 [-0.1081, -0.0179]	probability	conditional	brand2014 same scope as the hs_completion row; outcome = some college attendance by age 21-22, control mean 0.427 LANDED v1.42, same read as the hs_completion row: Table 3 kernel TT -0.063 (SE 0.023), p<.01; band = 1.96 SE. TIMING (Table 6 TT): 0-5 -0.032 n.s.; 6-11 -0.082* (SE 0.034); 12-17 -0.099* (SE 0.039). Beside: unmatched -0.089; TUT -0.099***. ESTIMAND DISCIPLINE: a stock (ever some college by ~21) - DISTINCT from the hilger2016 college_enrollment annual-flow node; the two must never be merged or converted. Conditional: same cohort/design transport as the hs_completion row
displacement event → college completion Full identifier `displacement_event->college_completion`	-0.036 [-0.0674, -0.0046]	probability	conditional	brand2014 same scope as the hs_completion row; outcome = college completion (BA) by age 25, control mean 0.169 LANDED v1.42, same read as the hs_completion row: Table 3 kernel TT -0.036 (SE 0.016), p<.05; band = 1.96 SE. TIMING (Table 6 TT): 0-5 -0.035 n.s.; 6-11 -0.027 n.s.; 12-17 -0.052 (SE 0.027, p<.10). Beside: unmatched -0.051; TUT -0.043*. The paper reports NO propensity trend for this outcome (unlike HS/attendance). Conditional: same cohort/design transport as the hs_completion row
displacement event → adult depression cesd Full identifier `displacement_event->adult_depression_cesd`	0.025 [0.0034, 0.0466]	scale01	conditional	brand2014 same scope as the hs_completion row; outcome = CESD 7-item depressive-symptom index (0-1) at ages 25-29, control mean 0.208 LANDED v1.42, same read as the hs_completion row: Table 3 kernel TT +0.025 (SE 0.011), p<.05 on the 0-1 index; band = 1.96 SE. Beside: CESD ages 20-24 +0.014 n.s. (SE 0.009); unmatched +0.030; TUT +0.034. TIMING (Table 6 TT): 0-5 -0.021 n.s.; 6-11 +0.047** (SE 0.017); 12-17 +0.034 (SE 0.019, p<.10); adolescence also raises CESD 20-24 (+0.031*, SE 0.015). CHANNEL: the displacement event transmits psychological distress to the CHILD into adulthood - the adult-child complement of spouse_mental_health_sd (partner) and unemployment_status->mental_health_sd (worker); direction-consistent, distinct unit (0-1 index, not SDs). Conditional: same cohort/design transport as the hs_completion row
displacement event → spouse participation elasticity Full identifier `displacement_event->spouse_participation_elasticity`	-0.04 [-0.07, -0.03]	log_elasticity	conditional	halla2020 married couples, Austrian ASOS registers 1990-2007, HUSBANDS displaced by firm closure/mass layoff vs non-event firm controls; wife's labor-force participation response LANDED v1.43 (sweep #3): Halla, Schmieder & Weber 2020 (AEJ:Applied 12(4):253-287, DOI 10.1257/app.20180671), Table 9 eta-participation + text: full-sample participation elasticity ~-0.04; subgroup range -0.03 (small-shock households) to -0.07 (previously-inactive wives, Table 6) - the BAND IS THE PAPER'S SUBGROUP RANGE, not a CI (dist empty). THE ADDED-WORKER EFFECT IS SMALL: one order of magnitude below the AWE-literature average (-0.4, the paper's own Table A1 survey); husbands lose 21-24% earnings / 16-17% employment over 5y while wives' employment rises ~1% (extensive margin only - NO intensive-margin response among already-employed wives); the response recovers a tiny fraction of the household loss; most responsive: mothers with children 3-15; zero response among childless wives/very young children; local labor markets condition it (no response when male unemployment above median). Read via IZA DP 11752 (WP draft of the AEJ version of record; read-via precedent). Conditional: Austrian institutions/UI generosity, married couples, closure/layoff identification
displacement event → annual birth response Full identifier `displacement_event->annual_birth_response`	-0.005 [-0.0089, -0.0011]	probability	conditional	huttunenkellokumpi2016 couples, Finnish FLEED registers, WOMEN displaced by plant closures (base years 1991-1993) vs matched non-displaced; outcome = couple's probability of a birth in the year LANDED v1.43 (sweep #4): Huttunen & Kellokumpi (Journal of Labor Economics 34(2), 2016, DOI 10.1086/683645 - VENUE CORRECTED from the sweep's 'Labour Economics'), IZA DP 6707 Table 3 'Gave birth' column, plant-closure sample: displacement-year marginal effect -0.005 (SE 0.002), p<.01; band = 1.96 SE on a ~10%/yr base (~5% relative reduction). Timing: year-3 -0.005** (0.002) recurs; years 2, 4-11 individually n.s. - the effect CUMULATES: ~-4 births per 100 displaced women by year 11 (Figure 6 event study; high-education women -0.05; low-education null - career-concern channel, NOT income: the couple's income drop does NOT predict the fertility response). MALE NULL declared beside (Table 4 'Gave birth': plant-closure dpl1 -0.002 (0.002) n.s.; mass-layoff -0.003*) - consistent with halla2020's husband-displacement births null (-0.001, SE 0.001). Distinct treatment from the shipped kearney2018fracking row (local-shock elasticity, never merged). Read via IZA DP 6707 (WP draft of the JLE version of record; read-via precedent). Conditional: Finnish registers, deep-recession displacement window (1991-93), women 20-40
displacement event → child depression anxiety Full identifier `displacement_event->child_depression_anxiety`	0.008 [0.0002, 0.0158]	probability	conditional	schallerzerpa2019 children ages 6-17, US MEPS, PATERNAL job loss (tenured-father sample, >=1y tenure), individual-FE LPM; outcome = child depression/anxiety report LANDED v1.43 (sweep #5): Schaller & Zerpa 2019 (American Journal of Health Economics; NBER w21745), Table 2 MENTAL HEALTH panel, father-displaced tenured column: +0.008pp (SE 0.004), p<.05; band = 1.96 SE on a small base (~100-150% relative increase per the authors). SPEC SENSITIVITY declared: firm-closure-only +0.006* (0.004), all-involuntary +0.002 n.s. - the effect concentrates where identification is strongest but imprecision grows. Beside: child mental-health-'excellent' report FALLS - father closure -0.044 (0.021), mother tenured -0.038 (0.017), mother closure -0.058 (0.026); physical health-'excellent' father closure -0.051* (0.017); MATERNAL depression/anxiety NULL declared (-0.004/-0.006/+0.003 all n.s.) - the paternal stress channel is the operative one, opposite to the fertility stream (maternal displacement operates there, huttunenkellokumpi2016) and the education stream (maternal, brand2014). THE CHILD'S OWN SHORT-RUN MENTAL HEALTH - distinct from adult outcomes (adult_depression_cesd ages 25-29, brand2014) and from the spouse's (marcus2013). HONESTY BESIDE (mork2019, Swedish workplace closures, registers, 10y): child hospitalization/mortality NULLS (maternal +1.8/1000 = +0.6% n.s.; paternal mental/behavior admissions -2.8/1000 = -8.3%, a DECLINE), paternal education NULLS; maternal GPA/HS negatives carry similar pre-trends and are questioned by the authors themselves - survey-vs-register outcome scope declared. Read via NBER w21745 (WP of the AJHE version of record; read-via precedent). Conditional: US MEPS survey reports, short-run window
child achievement sd → grandchild achievement sd Full identifier `child_achievement_sd->grandchild_achievement_sd`	0.38 [0.38, 0.42]	sd_delta	structural	blackdevereuxsalvanes2009 Norway male conscripts (IQ at age 18, father-son) father-son IQ correlation .38: 'The correlation we observe between father's IQ and that of his son is .38' (IZA DP 3651 p.8; identical to the Boucher & McGue 1981 weighted average). STANDARDIZED SLOPE: composes sd shifts linearly (linear_shift kind). CAVEAT carried from the authors: the correlation 'has no causal interpretation whatsoever' (p.9) - a displacement-induced shift is not assumed to transmit causally; this is a structural transmission assumption like the IGE band. Band pinned, not widened: lower edge = the single-measure replications (.38 here = .38 Boucher-McGue); upper edge = Bowles-Gintis 2002 lower bound .42 (their .42-.72 range uses two-parent averages, inflated per BDS p.8). IQ-at-18-to-achievement and Norway-to-US transport are declared caveats, not band widening. Read via IZA DP 3651 PDF (results section), NBER w14274 abstract beside
child education years → grandchild education years Full identifier `child_education_years->grandchild_education_years`	0.296 [0.255, 0.337]	edu_years	structural	lindahl2015 Sweden, Malmö four-generation register (G3 children, N=1,823) parent->child years-of-schooling slope 0.296 (SE 0.021), Table 2 cell (parent x child), bivariate regression with birth-year quadratics, family-clustered SEs; band = 95% CI (0.296 +/- 1.96 x 0.021), dist normal. LINEAR_SHIFT: composes year shifts linearly. CAVEAT carried from the authors: their own IV estimate (great-grandparent education as instrument) implies 'no causal effect of parental education on childrens' educational attainment' conditional on transmission (Sec. 4) - this row is a structural transmission assumption (persistence), not an identified intervention response. DANGLING UPSTREAM: no displacement->child_education_years entry yet (queue #33) - the hs_completion/college rows are probability-scale; the walk exists only once an admitted entry lands. Read via IZA DP 6463 full text (Table 2); published JHR 50(1) is the same study
infant birth weight → child earnings Full identifier `infant_birth_weight->child_earnings`	0.1 [0.0216, 0.1784]	gap_multiplier	conditional	black2007 Norway twin sample, twin fixed effects on ln(birth weight), cohorts 1967-1988 (military conscription + earnings registries); IZA DP 1864 Table 5 'Ln(Earnings) FT', all-twins FE column LANDED v1.45 fetal dose-response stream (user-directed 2026-09-14): coefficient on ln(BW) 0.10 (SE 0.04) - a ln-ln elasticity BY CONSTRUCTION: +10% BW -> +0.95% adult full-time earnings ('about 1%', authors' text); band = 1.96 x SE. OLS 0.10 (0.02) similar; all-earnings (incl. part-time) 0.09 (0.05); MZ-only column attenuates 0.09 (0.09) n.s. - recorded as declared upper-robustness, not the row. TRANSPORT/ROLE: a within-twin (fetal-growth) slope applied to a shock-induced BW shift is a structural_transmission_assumption - Almond-Currie 2011 caveat travels: BW is a crude fetal-health summary and the twin slope may not act identically along every margin. SCOPE: composes ONLY after displacement->infant_birth_weight (lindo2011, in-utero exposure window) - the IN-UTERO cohort; NEVER additive with displacement->child_earnings (oreopoulos2008 = TOTAL effect for the childhood-exposure cohort; different children). Composed point: 0.9541^0.10 = 0.995312 -> -0.47% adult earnings, ~an order of magnitude below the childhood channel (-9.24%).
infant birth weight → adult type2 diabetes hazard Full identifier `infant_birth_weight->adult_type2_diabetes_hazard`	-0.9637 [-1.3834, -0.5401]	odds_ratio	conditional	knop2018 adults, pooled random-effects dose-response, 32 studies / ~7.6M participants (Knop et al. JAHA 2018, 7(23):e008870, OR per +1 kg birth weight; read via PMC6405546 full text) LANDED v1.45 fetal dose-response stream: pooled OR 0.78 (95% CI 0.70-0.87) per +1 kg increment in birth weight (I2=72.6%). Log-elasticity conversion: ln(OR)/ln(4.4/3.4) with the US mean-birth-weight anchor 3.4 kg DECLARED (NCHS order-of-magnitude, canonical); band = CI endpoints converted identically. Sign convention: elasticity is negative (more BW, less diabetes risk); a displacement-induced BW LOSS composes to an odds INCREASE. Composed point: 0.9541^-0.9637 = 1.04632 -> +4.6% adult T2D odds on the in-utero cohort. SHAPE CAVEAT: the association is J-shaped with lowest risk at 3.5-4.0 kg - the inverse branch is the one a downward shift applies to; the high-tail upturn (>4.5 kg, female-concentrated) never enters this model's shifts and travels as undeclared heterogeneity (row is pooled-sex).
infant birth weight → adult cardiovascular disease hazard Full identifier `infant_birth_weight->adult_cardiovascular_disease_hazard`	-0.6994 [-0.8173, -0.585]	odds_ratio	conditional	knop2018 adults, pooled random-effects dose-response, 33 studies / ~5.9M participants (same source as the T2D row; OR per +1 kg birth weight) LANDED v1.45 fetal dose-response stream: pooled OR 0.835 (95% CI 0.81-0.86) per +1 kg increment in birth weight (I2=0.0 - the most homogeneous gradient in the meta-analysis). POINT PRECISION: the source sentence reads 'a 16.5% (OR: 0.84, 95% CI: 0.81-0.86) reduction' - the point is taken at its stated precision 1-0.165 = 0.835, which reconciles with the CI geometric center sqrt(0.81x0.86) = 0.8346 (the parenthetical 0.84 is 2-dp rounding). Same conversion and 3.4 kg declared anchor as the T2D row. Composed point: 0.9541^-0.6994 = 1.033419 -> +3.3% adult CVD odds on the in-utero cohort. Same J-shape caveat (lowest risk 4.0-4.5 kg; downward shifts stay on the inverse branch).
divorce hazard → child divorce hazard Full identifier `divorce_hazard->child_divorce_hazard`	1.88 [1.57, 2.24]	rate_ratio	structural	mclanahanbumpass1988 US women (NSFG cycles I-III), white daughters who spent part of childhood with a divorced/separated/never-married parent vs two-parent families; proportional-hazard model of own marital disruption LANDED v1.46 (sweep #6, queue row 35): McLanahan & Bumpass 1988 (AJS 94(1):130-152, DOI 10.1086/228954 verified via Crossref), Appendix Table A1 marital-disruption column + prose cross-check. Background-controlled model: beta 0.63 (SE 0.09); band = exp(beta +/- 1.96 SE) = 1.88 [1.57, 2.24]. ZERO-ORDER beta 0.65 (SE 0.09) = exp 1.92, matching the text's '92 percent more likely to experience a marital disruption' - the prose/table cross-check that validates the OCR read. BESIDE: education-controlled beta 0.51 -> HR 1.67 (the association survives own-education adjustment); Black women beta 0.31 (SE 0.08) -> HR 1.36 (smaller, same direction); widowhood weaker (beta 0.30) - disruption carries the effect, not parent absence per se. dist withheld: the band is exponentiated from a log-hazard CI, asymmetric around the multiplier (B&H precedent). READ-VIA: IRP DP 805-86 (WP of the AJS version of record; OCR pass - table digits validated against the stated prose percentages). THE WALK IS STILL BLOCKED: this row composes ONLY through a conditional/mixture kind (only the displaced marriages that actually dissolve transmit) - designed separately; gruber2004 is NOT a pin (law-exposure ITT, 'no rise in the odds of being divorced', venue corrected to JOLE 22(4))
married cohort → parental dissolution share Full identifier `married_cohort->parental_dissolution_share`	0.069 [0.069, 0.072]	probability	conditional	rege2007 married workers at Norwegian plants 1995-2000 (the rege2007 entry population and follow-up window, dissolution measured by 2003) LANDED v1.47 (queue row 35b): the counterfactual dissolution share for the conditional-mixture divorce transmission - SAME source, population, and window as the displacement->divorce_hazard entry (SSB DP 514 Table 1 + prose: 'About 7.0 percent of the couples got divorced from 1995 to 2003'; stable plants 6.9%, closing plants 7.2%). Point = the stable-plant share 0.069 (the counterfactual the walk compares against); band [0.069, 0.072] = the paper's observed range across plant types (no CI reported for the share). Feeds conditional_mixture: extra share = 1-(1-s)^m - s; at m=1.11, s=0.069 -> +0.73pp extra dissolutions, matching the paper's own adjusted +0.78pp (+11%) - the coherence check that the window and population align
undocumented share change pp → receiving violent crime log rate Full identifier `undocumented_share_change_pp->receiving_violent_crime_log_rate`	-0.12 [-0.18, -0.06]	log_delta	conditional	lightmiller2018 US 50 states + DC 1990-2014; lagged percentage-point undocumented population share; Light-Miller 2018 Table 2 Model 4 CORRECTED 2026-10-15 v1.52: published Criminology PDF SHA256 4af68d79d437943601cac72ba25692fbe999588e4fda88792d5729dfc0114df9, Table 1 p377 and focal-measure definition p379: undocumented share is unlogged, expressed in percentage points; violent crime rate is logged. Table 2 Model 4 p386: b=-0.12, clustered SE=0.03 (Models 3-5 have the same rounded b/SE). This is a semi-log slope per 1 percentage point, NOT a log-log immigrant-stock elasticity. The previous immigrant_influx linkage and +10% stock -> -1.2% claim were unsafe and retired. Source coefficient and rounded support [-0.18,-0.06] unchanged (1.96SE rounded to two decimals). Named adapter uses exp(bdelta_share_pp), annual outcome with predictors lagged one year, NOT a cumulative multi-year count. Preferred FE association; IV robustness does not turn this FE coefficient into the IV estimand. Population-weighted Model 6 b=-0.05 SE=.02 and cross-spec sensitivity remain beside. Not transferable to regular-resident or work-banned margins by relabeling.
regular immigrant share ratio → receiving total crime rate Full identifier `regular_immigrant_share_ratio->receiving_total_crime_rate`	0.105 [-0.2616, 0.4716]	gap_multiplier	conditional	bianchibuonannopinotti2012 Italy 95 provinces 1991-2001; log valid residence permits / total province population; 2010 working-paper ten-year-difference IV CORRECTED 2026-10-15 v1.52: source regressor is log(valid residence permits / total province population), not log immigrant stock alone. Accessible 2010 working-paper mirror SHA256 4857b7f55b25466dccb4df477191bdaee433d129c087fb68472a2884e2239daf, Eq3 p7 and Table5 col3 p33: IV b=.105, robust SE=.187, 95 Italian provinces, 1991-2001 differences. original rounded support [-.2616,.4716] unchanged and includes zero. Different from the formerly pinned HAL draft (SHA256 02442aa8b0c716be00cc598ba5948dce31e835ad2b19bd60cd1ea598a45e5c66, Table4 locator); published JEEA version not re-extracted in this correction. Tier canonical under CITING working-paper rule. Named adapter uses (post_share/reference_share)^b, including both total-population denominators. This is an endpoint crime RATE contrast, NOT ten-year cumulative incidents. Margin is valid residence-permit residents, not all immigrants, H-1B arrivals or asylum/work-banned migrants. OLS .156 SE.049 and alternative IV .029 SE.125 / -.029 SE.175 remain distinct specifications, never averaged or replaced by zero.

Baseline inventory

Cited absolute-rate and dollar inputs
Outcome	Value	Unit	Population, citation and derivation
all cause mortality annual	0.004944	deaths_per_person_year	US prime-age men (45-54) cdc_wonder D76 (Underlying Cause of Death, 1999-2020), Male 45-54 years, pooled 2015-2019: 514314 deaths / 104024440 person-years = 494.4 per 100k; pulled 2026-09-06 via scraper.cli wonder-mortality into warehouse wonder_mortality Pre-pandemic window: excludes 2020 (COVID-distorted; 2020 crude rate 601.5/100k). Single-year 2019 = 490.0/100k. Ages 45-54 match the Sullivan & von Wachter prime-age counterfactual population
divorce 5y cumulative	0.1045	prob_per_marriage	US married couples 5y window Census P70-125 (Kreider & Glick 2012, SIPP 2008 panel) Table 4: percent of first marriages reaching the 5th anniversary, marriage cohort 1995-1999 (newest with observable 5th anniversaries) — men 89.6 / women 89.5 percent -> 1 - 0.8955 = 0.1045; report prose: approximately 10 percent of first marriages had ended in divorce after 5 years; pulled 2026-09-06 via census-marriage-download into warehouse census_marriage_survival Survival semantics: percent NOT reaching the 5th anniversary intact (widowhood is also excluded from survival, and is small at 5 years). Separation-inclusive disruption is higher (NCHS NSFG reports about one fifth of first marriages disrupt within 5 years including separations) — this pin matches the model's divorce outcome, not separation
ipv annual incidence	0.0027	prob_per_person_year	US residents age 12 or older BJS Criminal Victimization 2024 (Tapp 2025, NCVS) violent-crime trends table: intimate partner violence rate 2.7 per 1,000 persons age 12+ in 2024 (2020-2024 window: 1.7/1.7/3.4/2.2/2.7); pulled 2026-09-06 via bjs-ipv-download into warehouse bjs_ncvs_victimization UNIT CORRECTED at pin time: the placeholder row declared a per-household rate, but NCVS publishes per-person rates — pinned semantics follow the publication. Victimization counts (a person with multiple incidents counts per incident), not unique victims. The 2022 rate (3.4) is the window high; pin uses the most recent year (2024). NCVS undercounts relative to self-administered surveys for sensitive victimizations — documented NCVS limitation
youth crime participation	0.055948	prob_per_youth_year	US young men 18-24 OJJDP SBB arrest trends (api.ojp.gov dataset wwuj-iznp; NCJJ estimates from FBI Crime in the United States): male arrests age 18-20 (360,130) + 21-24 (548,750) = 908,880 in 2023; denominator Census Population Estimates nc-est2025-agesex-res.csv male 18-24 2023 = 16,245,172; 908,880/16,245,172 = 0.055948 (5,595 arrests per 100,000); both pulls 2026-09-06 via ojjdp-arrests-download + census-popest-download into warehouse ojjdp_arrests + census_popest_agesex BRACKET CORRECTION at pin time: the placeholder population said 16-24; the published brackets are 18-20 and 21-24, so the pin covers 18-24 (16-17-year-olds sit in the juvenile system whose counts are a separate SBB series). Arrest counts, not unique arrestees (a person arrested twice counts twice) — this is the plan's declared arrests-per-100k metric, an upper bound on first-arrest participation
median male lifetime earnings	2591418.0	usd_2024	US male median lifetime earnings SSA Annual Statistical Supplement 2025, Table 4.B6 (CWHS/Master Earnings File, 1 percent sample): male median earnings by age, 2023 (preliminary) — synthetic career ages 20-64 = 5x(24218+42561+52485+60423+64962+67506+68160+64854)+2x60666+3x56666 = 2517175 usd_2023; CPI-U annual averages 304.702->313.689 (BLS series CUUR0000SA0, pulled 2026-09-06) -> 2591418 usd_2024; warehouse ssa_median_earnings (pulled via ssa-earnings-download) Derivation declared: the cross-sectional age profile summed over a 45-year working life (ages 20-64) is a synthetic cohort — the sum of age-band medians approximates rather than equals the median of individual lifetime sums; median-of-workers semantics excludes non-working years; 2023 medians are preliminary (SSA replaces them with final data in a later edition)

Source bibliography

Inclusion in the bibliography does not establish peer review or independent validation.

adermon2018: Adermon, Adrian; Lindahl, Mikael; Waldenström, Daniel. Intergenerational Mobility in Pension Income: An Adoptee Perspective (2018). Economica.
adh2013: Autor, David H.; Dorn, David; Hanson, Gordon H.. The China Syndrome: Local Labor Market Effects of Import Competition in the United States (2013). American Economic Review.
aizer2010: Aizer, Anna. The Gender Wage Gap and Domestic Violence (2010). American Economic Review.
akee2010: Akee, Randall K. Q.; Copeland, William E.; Keeler, Gordon; Angold, Adrian; Costello, E. Jane. Parents' Incomes and Children's Outcomes: A Quasi-Experiment Using Transfer Payments from Casino Profits (2010). American Economic Journal: Applied Economics.
alesina1999: Alesina, Alberto; Baqir, Reza; Easterly, William. Public Goods and Ethnic Divisions (1999). Quarterly Journal of Economics.
ananat2017: Ananat, Elizabeth O.; Gassman-Pines, Anna; Francis, David V.; Gibson-Davis, Christina M.. Linking Job Loss, Inequality, Mental Health, and Education (2017). Science.
autor2003: Autor, David H.; Duggan, Mark G.. The Rise in the Disability Rolls and the Decline in Unemployment (2003). Quarterly Journal of Economics.
autor2019: Autor, David; Dorn, David; Hanson, Gordon. When Work Disappears: Manufacturing Decline and the Falling Marriage-Market Value of Young Men (2019). American Economic Review: Insights.
autor2020: Autor, David; Dorn, David; Hanson, Gordon; Majlesi, Kaveh. Importing Political Polarization? The Electoral Consequences of Rising Trade Exposure (2020). American Economic Review.
autor2020polarization: Autor, David; Dorn, David; Hanson, Gordon; Majlesi, Kaveh. Importing Political Polarization? The Electoral Consequences of Rising Trade Exposure (2020). American Economic Review.
bastian2018: Bastian, Jacob; Michelmore, Katherine. The Long-Term Impact of the Earned Income Tax Credit on Children's Education and Employment Outcomes (2018). Journal of Labor Economics.
bellfasanimachin2013: Bell, Brian; Fasani, Francesco; Machin, Stephen. Crime and Immigration: Evidence from Large Immigrant Waves (2013). Review of Economics and Statistics.
bianchibuonannopinotti2012: Bianchi, Milo; Buonanno, Paolo; Pinotti, Paolo. Do Immigrants Cause Crime? (2012). Journal of the European Economic Association.
bingley2026: Bingley, Paul; Cappellari, Lorenzo; Ovidi, Marco. When It Hurts the Most: Timing of Parental Job Loss and a Child's Education (2026). Journal of the European Economic Association.
black2003: Black, Dan A.; McKinnish, Terra G.; Sanders, Seth G.. Does the Availability of High-Wage Jobs for Low-Skilled Men Affect Welfare Expenditures? Evidence from Shocks to the Coal and Steel Industries (2003). Review of Economics and Statistics.
black2005: Black, Sandra E.; Devereux, Paul J.; Salvanes, Kjell G.. The More the Merrier? The Effect of Family Size and Birth Order on Children's Education (2005). Quarterly Journal of Economics.
black2007: Black, Sandra E.; Devereux, Paul J.; Salvanes, Kjell G.. From the Cradle to the Labor Market? The Effect of Birth Weight on Adult Outcomes (2007). Quarterly Journal of Economics.
blackdevereuxsalvanes2009: Black, Sandra E.; Devereux, Paul J.; Salvanes, Kjell G.. Like Father, Like Son? A Note on the Intergenerational Transmission of IQ Scores (2009). Economics Letters.
bloemen2018: Bloemen, Hans; Stancanelli, Elena; van der Klaauw, Bas. Job Loss, Firm-Level Heterogeneity and Mortality: Evidence from Administrative Data (2018). Journal of Health Economics.
borjas2003: Borjas, George J.. The Labor Demand Curve Is Downward Sloping: Reexamining the Impact of Immigration on the Labor Market (2003). Quarterly Journal of Economics.
brand2014: Brand, Jennie E.; Simon Thomas, Juli. Job Displacement Among Single Mothers: Effects on Children's Outcomes in Young Adulthood (2014). American Journal of Sociology.
browning2006: Browning, Martin; Dano, Anne Møller; Heinesen, Eskil. Job Displacement and Stress: Evidence from Labor-Market Displacement in Denmark (2006). Journal of Health Economics.
browningheinesen2012: Browning, Martin; Heinesen, Eskil. Effect of Job Loss Due to Plant Closure on Mortality and Hospitalization (2012). Journal of Health Economics.
butcher1998: Butcher, Kristin F.; Piehl, Anne Morrison. Recent Immigrants: Unexpected Implications for Crime and Incarceration (1998). Industrial and Labor Relations Review.
campbell2011: Campbell, John Y.; Giglio, Stefano; Pathak, Parag. Forced Sales and House Prices (2011). American Economic Review.
card1990: Card, David. The Impact of the Mariel Boatlift on the Miami Labor Market (1990). Industrial and Labor Relations Review.
case2015: Case, Anne; Deaton, Angus. Rising Morbidity and Mortality in Midlife among White Non-Hispanic Americans in the 21st Century (2015). PNAS.
charles2004: Charles, Kerwin Kofi; Stephens, Melvin, Jr.. Job Displacement, Disability, and Divorce (2004). Journal of Labor Economics.
chetty2014: Chetty, Raj; Hendren, Nathaniel; Kline, Patrick; Saez, Emmanuel. Where is the Land of Opportunity? The Geography of Intergenerational Mobility in the United States (2014). Quarterly Journal of Economics.
chetty2016: Chetty, Raj; Stepner, Michael; Abraham, Sarah; Lin, Shelby; Scuderi, Benjamin; Turner, Nicholas; Bergeron, Augustin; Cutler, David. The Association Between Income and Life Expectancy in the United States, 2001--2014 (2016). JAMA.
chetty2016jama: Chetty, Raj; Stepner, Michael; Abraham, Sarah; Lin, Shelby; Scuderi, Benjamin; Turner, Nicholas; Schrag, Augustin; Cutler, David. The Association Between Income and Life Expectancy in the United States, 2001--2014 (2016). JAMA.
chettyhendren2018: Chetty, Raj; Hendren, Nathaniel. The Impacts of Neighborhoods on Intergenerational Mobility I: Childhood Exposure Effects (2018). The Quarterly Journal of Economics.
colantone2018: Colantone, Italo; Stanig, Piero. The Trade Origins of Economic Nationalism: Import Competition and Voting Behavior in Western Europe (2018). American Journal of Political Science.
collinson2024: Collinson, Robert; Humphries, John Eric; Mader, Nicholas S.; Reed, Davin K.; Tannenbaum, Daniel I.; van Dijk, Winnie. Eviction and Poverty in American Cities (2024). The Quarterly Journal of Economics.
corak2013: Corak, Miles. Income Inequality, Equality of Opportunity, and Intergenerational Mobility (2013). Journal of Economic Perspectives.
cortes2016: Cortes, Guido Matias. Where Have the Middle-Wage Workers Gone? A Study of Polarization Using Panel Data (2016). Journal of Labor Economics.
costello2003: Costello, E. Jane; Compton, Stephen N.; Keeler, Gordon; Angold, Adrian. Relationships Between Poverty and Psychopathology: A Natural Experiment (2003). JAMA.
culhane2007: Culhane, Dennis P.; Metraux, Stephen; Hadley, Thomas. Public Service Reductions Associated with Placement of Homeless Persons with Severe Mental Illness in Supportive Housing (2007). Housing Policy Debate.
currie2003: Currie, Janet; Moretti, Enrico. Mother's Education and the Intergenerational Transmission of Human Capital: Evidence from College Openings (2003). Quarterly Journal of Economics.
dahl2012: Dahl, Gordon B.; Lochner, Lance. The Impact of Family Income on Child Achievement: Evidence from the Earned Income Tax Credit (2012). American Economic Review.
damm2014: Damm, Anna Piil; Dustmann, Christian. Does Growing Up in a High Crime Neighborhood Affect Youth Criminal Behavior? (2014). American Economic Review.
davis2011: Davis, Steven J.; von Wachter, Till. Recessions and the Costs of Job Loss (2011). Brookings Papers on Economic Activity.
desmond2016: Desmond, Matthew; Gershenson, Carl. Housing and Employment Insecurity among the Working Poor (2016). Social Problems.
doran2022: Doran, Kirk; Gelber, Alexander; Isen, Adam. The Effects of High-Skilled Immigration Policy on Firms: Evidence from Visa Lotteries (2022). Journal of Political Economy Microeconomics.
duncan2010: Duncan, Greg J.; Ziol-Guest, Kathleen M.; Kalil, Ariel. Early-Childhood Poverty and Adult Attainment, Behavior, and Health (2010). Child Development.
ehrensaft2003: Ehrensaft, Miriam K.; Cohen, Patricia; Brown, Jocelyn; Smailes, Elizabeth; Chen, Henian; Johnson, Jeffrey G.. Intergenerational Transmission of Partner Violence: A 20-Year Prospective Study (2003). Journal of Consulting and Clinical Psychology.
eliason2009: Eliason, Marcus; Storrie, Donald. Does Job Loss Shorten Life? (2009). Journal of Human Resources.
eliasonstorrie2009: Eliason, Marcus; Storrie, Donald. Does Job Loss Shorten Life? (2009). Journal of Human Resources.
evans2008: Evans, Sarah E.; Davies, Corrie; DiLillo, David. Exposure to Domestic Violence: A Meta-Analysis of Child and Adolescent Outcomes (2008). Aggression and Violent Behavior.
farrington2006: Farrington, David P.; Coid, Jeremy W.; Harnett, Lucy M.; Jolliffe, Darrick; Soteriou, Nicola; Turner, Rebecca; West, Donald J.. Criminal Careers up to Age 50 and Life Success up to Age 48: New Findings from the Cambridge Study in Delinquent Development (2006). Home Office Research Study 299.
fetzer2019: Fetzer, Thiemo. Did Austerity Cause Brexit? (2019). American Economic Review.
galofrevila2021: Galofré-Vilà, Gregori; Meissner, Christopher M.; McKee, Martin; Stuckler, David. Austerity and the Rise of the Nazi Party (2021). Journal of Economic History.
ganong2022: Ganong, Peter; Noel, Pascal. Liquidity versus Wealth in Household Debt Obligations: Evidence from Housing Policy in the Great Recession (2022). Quarterly Journal of Economics.
gigerenzer2002: Gigerenzer, Gerd. Calculated Risks: How to Know When Numbers Deceive You (2002). Simon & Schuster.
gould2002: Gould, Eric D.; Weinberg, Bruce A.; Mustard, David B.. Crime Rates and Local Labor Market Opportunities in the United States: 1979--1997 (2002). Review of Economics and Statistics.
grogger1998: Grogger, Jeff. Market Wages and Youth Crime (1998). Journal of Labor Economics.
gruber2004: Gruber, Jonathan. Is Making Divorce Easier Bad for Children? The Long-Run Implications of Unilateral Divorce (2004). Journal of Labor Economics.
halla2020: Halla, Martin; Schmieder, Johannes; Weber, Andrea. Job Displacement, Family Dynamics, and Spousal Labor Supply (2020). American Economic Journal: Applied Economics.
hersbach2000: Hersbach, Hans. Decomposition of the Continuous Ranked Probability Score for Ensemble Prediction Systems (2000). Weather and Forecasting.
hilger2016: Hilger, Nathaniel G.. Parental Job Loss and Children's Long-Term Outcomes: Evidence from 7 Million Fathers' Layoffs (2016). American Economic Journal: Applied Economics.
holliingsworth2017: Hollingsworth, Alex; Ruhm, Christopher J.; Simon, Kosali. Macroeconomic Conditions and Opioid Abuse (2017). Journal of Health Economics.
huttunen2019: Huttunen, Kristiina; Riukula, Krista. Parental Job Loss and Children's Careers (2019). IZA Discussion Paper 12788.
huttunenkellokumpi2016: Huttunen, Kristiina; Kellokumpi, Jenni. The Effect of Job Displacement on Couples' Fertility Decisions (2016). Journal of Labor Economics.
iman1982: Iman, Ronald L.; Conover, W. J.. A Distribution-Free Approach to Inducing Rank Correlation Among Input Variables (1982). Communications in Statistics - Simulation and Computation.
immergluck2006: Immergluck, Dan; Smith, Geoff. The Impact of Single-Family Mortgage Foreclosures on Neighborhood Crime (2006). Housing Studies.
jackson2016: Jackson, C. Kirabo; Johnson, Rucker C.; Persico, Claudia. The Effects of School Spending on Educational and Economic Outcomes: Evidence from School Finance Reforms (2016). Quarterly Journal of Economics.
jacobson1993: Jacobson, Louis S.; LaLonde, Robert J.; Sullivan, Daniel G.. Earnings Losses of Displaced Workers (1993). American Economic Review.
kearney2018fracking: Kearney, Melissa S.; Wilson, Riley. Male Earnings, Marriageable Men, and Nonmarital Fertility: Evidence from the Fracking Boom (2018). Review of Economics and Statistics.
kearney2020: Kearney, Melissa S.; Wilson, Riley. Male Earnings, Marriageable Men, and Nonmarital Fertility: Evidence from the Birth of the Pill (2020). Review of Economics and Statistics.
knop2018: Knop, Michael R.; Geng, Tianti T.; Gorny, Anna W.; Ding, Rui; Li, Changwei; Ley, Sylvia H.; Huang, Tao. Birth Weight and Risk of Type 2 Diabetes Mellitus, Cardiovascular Disease, and Hypertension in Adults: A Meta-Analysis of 7 646 267 Participants From 135 Studies (2018). Journal of the American Heart Association.
kposowa2001: Kposowa, Augustine J.. Unemployment and Suicide: A Cohort Analysis of Social Factors Predicting Suicide in the US National Longitudinal Mortality Study (2001). Psychological Medicine.
lightheobey2020: Light, Michael T.; He, Jingying; Robey, Jason P.. Comparing Crime Rates between Undocumented Immigrants, Legal Immigrants, and Native-Born US Citizens in Texas (2020). Proceedings of the National Academy of Sciences.
lightmiller2018: Light, Michael T.; Miller, Ty. Does Undocumented Immigration Increase Violent Crime? (2018). Criminology.
lindahl2015: Lindahl, Mikael; Palme, Mårten; Massih, Sofía Sandgren; Sjögren, Anna. Long-Term Intergenerational Persistence of Human Capital: An Empirical Analysis of Four Generations (2015). Journal of Human Resources.
lindo2011: Lindo, Jason M.. Parental Job Loss and Infant Health (2011). Journal of Health Economics.
lindo2018: Lindo, Jason M.; Schaller, Jessamyn; Hansen, Benjamin. Causes and Consequences of Child Maltreatment: Evidence from the Supplemental Nutrition Assistance Program (2018). American Economic Journal: Applied Economics.
lindo2018maltreatment: Lindo, Jason M.; Schaller, Jessamyn; Hansen, Benjamin. Caution! Men Not at Work: Gender-Specific Labor Market Conditions and Child Maltreatment (2018). Journal of Public Economics.
luttmer2001: Luttmer, Erzo F. P.. Group Loyalty and the Taste for Redistribution (2001). Journal of Political Economy.
marcus2013: Marcus, Jan. The Effect of Unemployment on the Mental Health of Spouses: Evidence from Plant Closures in Germany (2013). Journal of Health Economics.
mckay1979: McKay, Michael D.; Beckman, Richard J.; Conover, William J.. A Comparison of Three Methods for Selecting Values of Input Variables in the Analysis of Output from a Computer Code (1979). Technometrics.
mclanahan1994: McLanahan, Sara; Sandefur, Gary. Growing Up with a Single Parent: What Hurts, What Helps (1994). Harvard University Press.
mclanahanbumpass1988: McLanahan, Sara; Bumpass, Larry. Intergenerational Consequences of Family Disruption (1988). American Journal of Sociology.
moretti2010: Moretti, Enrico. Local Multipliers (2010). American Economic Review.
mork2019: Mörk, Eva; Sjögren, Anna; Svaleryd, Helena. Consequences of Parental Job Loss on the Family Environment and on Human Capital Formation: Evidence from Plant Closures (2019). IZA Discussion Paper 12559.
naidich2020: National Academies of Sciences, Engineering, and Medicine. The Economic and Fiscal Consequences of Immigration (2017). National Academies Press.
oreopoulos2008: Oreopoulos, Philip; Page, Marianne E.; Stevens, Ann Huff. The Intergenerational Effects of Worker Displacement (2008). Journal of Labor Economics.
oreopoulos2012: Oreopoulos, Philip; von Wachter, Till; Heisz, Andrew. The Short- and Long-Term Career Effects of Graduating in a Recession (2012). American Economic Journal: Applied Economics.
ottaviano2012: Ottaviano, Gianmarco I. P.; Peri, Giovanni. Rethinking the Effect of Immigration on Wages (2012). Journal of the European Economic Association.
ousey2018: Ousey, Graham C.; Kubrin, Charis E.. Immigration and Crime: Assessing a Contentious Issue (2018). The Annals of the American Academy of Political and Social Science.
paul2009: Paul, Karsten I.; Moser, Klaus. Unemployment Impairs Mental Health: Meta-Analyses (2009). Journal of Vocational Behavior.
pettigrew2006: Pettigrew, Thomas F.; Tropp, Linda R.. A Meta-Analytic Test of Intergroup Contact Theory (2006). Journal of Personality and Social Psychology.
putnam2007: Putnam, Robert D.. E Pluribus Unum: Diversity and Community in the Twenty-first Century (2007). Scandinavian Political Studies.
raphael2001: Raphael, Steven; Winter-Ebmer, Rudolf. Identifying the Effect of Unemployment on Crime (2001). Journal of Law and Economics.
rege2007: Rege, Mari; Telle, Kjetil; Votruba, Mark. Plant Closure and Marital Dissolution (2007). Statistics Norway Discussion Paper No. 514.
royer2009: Royer, Heather. Separated at Girth: US Twin Estimates of the Effects of Birth Weight (2009). American Economic Journal: Applied Economics.
saltelli2002: Saltelli, Andrea. Making Best Use of Model Evaluations to Compute Sensitivity Indices (2002). Computer Physics Communications.
schallerstevens2015: Schaller, Jessamyn; Stevens, Ann Huff. Short-Run Effects of Job Loss on Health Conditions, Health Insurance, and Health Care Utilization (2015). Journal of Health Economics.
schallerzerpa2019: Schaller, Jessamyn; Zerpa, Mariana. Short-Run Effects of Parental Job Loss on Child Health (2019). American Journal of Health Economics.
schneider2016: Schneider, Daniel; Harknett, Kristen; McLanahan, Sara. Intimate Partner Violence in the Great Recession (2016). Demography.
sobol2001: Sobol', Ilya M.. Global Sensitivity Indices for Nonlinear Mathematical Models and Their Monte Carlo Estimates (2001). Mathematics and Computers in Simulation.
solon1992: Solon, Gary. Intergenerational Income Mobility in the United States (1992). American Economic Review.
stevens2011: Stevens, Ann Huff; Schaller, Jessamyn. Short-Run Effects of Parental Job Loss on Children's Academic Achievement (2011). Economics of Education Review.
sullivan2000: Sullivan, Teresa A.; Warren, Elizabeth; Westbrook, Jay Lawrence. The Fragile Middle Class: Americans in Debt (2000). Yale University Press.
sullivan2009: Sullivan, Daniel; von Wachter, Till. Job Displacement and Mortality: An Analysis Using Administrative Data (2009). Quarterly Journal of Economics.
vonwachter2009: von Wachter, Till; Song, Jae; Manchester, Joyce. Long-Term Earnings Losses due to Job Separation During the 1982 Recession: An Empirical Analysis (2009). Journal of Labor Economics.
widom1989: Widom, Cathy Spatz. The Cycle of Violence (1989). Science.
worldbank2023: World Bank. Migration and Development Brief 39 (2023). World Bank Migration and Development Brief.
yagan2019: Yagan, Danny. Employment Hysteresis from the Great Recession (2019). Journal of Political Economy.
yang2008: Yang, Dean. International Migration, Remittances and Household Investment: Evidence from Philippine Migrants' Exchange Rate Shocks (2008). Economic Journal.
zg2012: Ziol-Guest, Kathleen M.; Duncan, Greg J.; Kalil, Ariel; Boyce, W. Thomas. Early Childhood Poverty, Immune-Mediated Disease Processes, and Adult Productivity (2012). Proceedings of the National Academy of Sciences.