Long-run human-capital effects of childhood forced displacement
LLM status Open unverified
J24 — Human Capital - Skills - Occupational Choice - Labor Productivity Economics catalogue openaigpt 6 astra ultra 45% cataloguednot recordedOpenStatement
ECON-076 | R | Source-backed agenda; exact formalization is editorial
Definitions and formal target
At a common pre-policy date $t_0$, fix a child cohort and baseline age $a_i$. Two specified regimes $z_1,z_0$ govern displacement timing, duration, destination and support. Potential adult schooling, earnings and health are $S_i(z),Y_i(z),H_i(z)$, measured at a common age or horizon. Target \[ \theta(a)=\mathbb E[(S_i(z_1)-S_i(z_0),\,Y_i(z_1)-Y_i(z_0),\, H_i(z_1)-H_i(z_0))\mid a_i=a]. \] Identify or bound persistence and heterogeneity, accounting for family separation, selective movement, mortality and loss to follow-up.
Clarifications and required assumptions
A child's own childhood-to-adulthood outcome is long-run persistence, not by itself an intergenerational effect. Age is defined at a common baseline, rather than actual displacement age, which can be changed by policy and undefined for never-displaced children. State how regimes differ and which legal-status/support components are bundled. Outcomes after death require a convention or joint survival analysis; selective tracing is not ignorable by definition.
Variants and changes of scope
- Age-specific support effects among already-displaced baseline children differ from causal effects of displacement itself.
- Effects on these children's descendants are a genuinely intergenerational extension requiring a new cohort/fertility definition and separate evidence.
- Different durations, destinations and reunification policies are distinct interventions, not interchangeable treatment versions.
Audit conclusion and source relationship
[S1] and [S2] support long-run outcomes of children and forced displacement. The original intergenerational title did not match its same-child adult outcome; title and age conditioning are corrected. The descendant extension is editorial and separately labeled.
References and exact source roles
[S1] Sandra V. Rozo and Guy Grossman (2025). Refugees and Other Forcibly Displaced Populations. VoxDevLit 14(1), March 2025.
Verified locator / source role: Conclusion/evidence-gap chapter at the linked primary review URL, inspected for the agenda. See the audit conclusion for distinctions between direct agenda support and editorial extensions.
[S2] Oliver Hanney / VoxDev (living compilation). Evidence gaps: Key unanswered questions in development economics. Accessed 8 October 2026. The page currently covers 20 topics; the "16-topics URL is a historical slug.
Verified locator / source role: Topic heading: Refugees and Other Forcibly Displaced Populations. Supports the broad research agenda; this is not a verbatim quotation or an attribution of the displayed mathematical target.
Common conventions: Additional-source status and mathematical conventions
Of the 100 supplied ECON entries, 65 distinct problems appear here; 32 close matches refine earlier entries, and three duplicate questions map to existing entries. Appendix app:audited-crosswalk lists every destination. The attachment's P/R/S/U distinctions and source audits are retained. Its precise mathematical formalizations are editorial unless the individual audit says otherwise. Candidate subsidy targets ECON-004--006 retain their U flags; the resolved approval-core question ECON-007 updates OP-0202 above. The following supplied conventions govern both this part and the attributed ECON refinements elsewhere in the catalogue, subject to local restrictions.
Parameterized questions. A problem family lists inputs that must be fixed before an instance is evaluated: population, horizon, policies, information, data law, model class and welfare weights. These are required choices, not quantities the citations are assumed to specify. Undefined applications should not be treated as identified estimands.
Indices and integrability. Sets of agents, firms and regions are finite unless a population measure is explicitly used. $i,j$ index units, $r$ regions, $t$ dates and $h$ horizons. $\mathbb E$ and $\Pr$ refer to the declared population and law; $\mathbf1\{A\}$ is an indicator. Every displayed expectation and discounted sum needs existence in the stated domain. Infinite horizons use a discount in $(0,1)$ and a convergence condition; finite horizons may use discount one.
Optimization and existence. A displayed $\max$, $\min$, or $\arg\max$ is conditional on attainment. For finite feasible menus this is immediate; general spaces need continuity and compactness/coercivity or another existence argument. Without attainment, the target is the supremum/infimum and arbitrarily near-optimal feasible choices. $\inf\varnothing=+\infty$. Empty feasibility and an unidentified target are different issues.
Potential outcomes. $Y_i(z)$ is the outcome under a specified intervention, not the observed conditional mean among people choosing $z$. In binary comparisons $\Delta Y_i=Y_i(1)-Y_i(0)$. Specify assignment, treatment versions, compliance, information and observation horizon. Interference is allowed under a full policy vector; compressed exposure notation needs a justified mapping. No interference, consistency and overlap are substantive assumptions, not automatic consequences of notation.
Populations and observation. Fix baseline eligibility and population weights. Conditioning on post-policy employment, survival, relocation or program uptake can change the population being compared. Death is not ordinary loss to follow-up; outcomes truncated by death need a composite convention, joint survival target, or explicitly defined survivor stratum. Fertility-changing interventions need a population functional rather than an unexplained fixed descendant cohort. Measurement and missingness models must be supplied.
Identification. A structural model $m\in\mathcal M$ implies observable law $P_m$ and target $\theta(m)$. Identification means \[ P_{m_1}=P_{m_2}\ \Longrightarrow\ \theta(m_1)=\theta(m_2) \quad\text{for all }m_1,m_2\in\mathcal M . \] Without identification, the target is \[ \Theta(P;\mathcal M)=\{\theta(m):m\in\mathcal M,\ P_m=P\}. \] Writing $\theta(P)$ before establishing identification can hide incompatible latent models. Confidence sets additionally need a sampling law, confidence level, asymptotic sequence and uniformity class. Point identification, coverage and informative precision are separate properties.
Mediation. Total effects of randomized assignment are distinct from cross-world natural indirect effects. Belief/effort-channel effects require a meaningful mediator intervention and additional measurement, exclusion or mediation assumptions. Randomization of the initial policy alone does not supply those assumptions. Report bounds or interventional alternatives when the stronger target is unsupported.
Equilibrium. A counterfactual $W(z)$, $p(z)$, or $\mu_t^z$ requires individual optimization, feasibility, government/resource budgets, market clearing and state-transition laws. State equilibrium existence and selection, or report ranges over compatible equilibria. Initial-state integration includes subsequent endogenous transitions; current-state integration must use the policy-dependent distribution. Reduced-form invariance after policy is not assumed.
Welfare accounts. Scores, utility, health, dollars and ecological quantities require explicit conversion before addition. Fees, taxes, subsidies, wages and extortion payments are transfers between parties; distributional weights can make them welfare relevant, but their face value is not automatically a resource loss. State ownership, geographic boundaries, foreign-income weights and fiscal finance. Do not add profits to household welfare already including dividends, or count land capitalization and the underlying benefit twice. A benefit-cost expression must distinguish gross benefits from net welfare and subtract every real cost exactly once.
Derivatives and risk. Log derivatives require positive arguments; local responses require admissible perturbations and specified equilibrium branches. Differentiation under expectations needs regularity/domination, and boundaries may allow only one-sided derivatives. Risk uses a specified law; ambiguity uses a declared nonempty law set on a common history space. Adaptive policies must be nonanticipative. A robust criterion is an editorial choice unless attributed explicitly.
Scope overlaps. Allocation, collective choice, contracts, household bargaining and coordinated policy overlap economics with optimization and game theory. They are retained because they appeared in the original catalogue; no claim of a strictly game-theory-free selection is made.
Statement licence: CC-BY-4.0 · Source file
Contributions 1
GPT 6 Astra Ultra attempt on Econ-519
Childhood displacement: age profiles, tracing, and survivor ambiguity
Problem ID:J24-515. Status: Unresolved. The full catalogue target remains unresolved; the results below are restricted theoretical contributions.The cohort target
Fix children already born at the common baseline, their baseline age $a$, and two complete regimes $z=0,1$ governing displacement timing, duration, destination and support. The catalogue asks for adult schooling, earnings and health differences at a common adult age. Neither conditioning on those who eventually migrate nor comparing children by their realized age at displacement preserves this target. Those variables can change with the regime.The displacement research agenda in Rozo and Grossman's review motivates the long horizon and difficulty of following mobile populations. The results below concern a restricted randomized-regime benchmark and transparent stock dynamics. They are not estimates of any displacement event. Randomization of support among already displaced children, although potentially informative, identifies a support-policy contrast and cannot by itself identify displacement versus no displacement.
A human-capital stock calculation
Let chronological age be $t$, the common adult measurement age be $A$, and human capital satisfy \[ h_{t+1}(z)=\rho h_t(z)+b_t-q_t(z),\qquad 0<\rho\le1. \] The deterministic $b_t$ denotes common investment and $q_t$ a regime-dependent net disruption, allowing negative values if additional support more than offsets lost inputs. At common baseline age $a$ the stock is the same under both regimes. Repeated substitution gives the exact identity \[ h_A(1)-h_A(0)=-\sum_{t=a}^{A-1}\rho^{A-1-t}[q_t(1)-q_t(0)]. \tag{1} \]A numerical check uses $\rho=1/2$, $A=5$, $L=1$, $d=8$. A shock at age $1$ lowers $h_5$ by $1$, while a shock at age $3$ lowers it by $4$. These are deliberately artificial numbers. Their role is to isolate a mathematical possibility, not to deny empirical evidence of sensitive periods. Mapping $h_A$ into earnings additionally requires wage prices, skills demand and labor supply; common schooling attainment is not a sufficient statistic for those mechanisms.
Randomized regimes with incomplete follow-up
For one adult outcome with a fixed convention for death, write $Y(z)\in[L,U]$. Let $Z$ be randomized with positive probabilities and $R(z)$ indicate initial successful follow-up. Consistency gives observed $(R,RY)$ in each arm. Put $r_z={\mathbb{P}}(R=1\mid Z=z)$ and $m_z=E[Y\mid R=1,Z=z]$, interpreting $r_zm_z$ as $E[RY\mid Z=z]$ when $r_z=0$.Probability tracing recovers the missing component
Suppose among initial nonrespondents a random tracing sample is selected with known probability $\pi_z(X)>0$, where $X$ includes baseline location and age, and the tracing procedure succeeds in observing the defined outcome for every selected child. Let $T$ indicate tracing selection, zero for original respondents. Conditional on the original nonrespondent population, require selection to be independent of the outcome given $(X,Z)$. Then \[ \mu_z=E\left[RY+(1-R)\frac{TY}{\pi_z(X)}\ \middle|\ Z=z\right]. \tag{2} \] Indeed conditional expectation of $TY/\pi_z(X)$ given $(Y,X,R=0,Z=z)$ equals $Y$. Integrating proves (2) without assuming the original nonresponse was random. For finite variance one additionally needs an integrable squared outcome divided by the tracing probability; probabilities bounded away from zero and bounded outcomes suffice. The resulting estimator can be inefficient if hard-to-find children receive extremely small probabilities.If selected children still cannot be found, the same argument only identifies outcomes of those successfully traced. One must then randomize a further successful ascertainment mechanism, impose explicit selection assumptions, or retain bounds for the residual missing fraction. A tracing attempt is not itself outcome observation. This distinction is especially consequential when survival status and international movements are incompletely recorded.
Death and the survivor estimand
A composite outcome that assigns a stated value to death is defined for the original cohort, but its interpretation depends on that convention. Alternatively let $S(z)$ indicate survival to the adult horizon and seek the always-survivor effect $E[Y(1)-Y(0)\mid S(1)=S(0)=1]$. Even complete randomized follow-up generally does not identify this cross-world population.Here is a concrete demonstration. In both arms half survive and survivor outcomes are equally likely $0$ or $1$. In model A the same half survive under both regimes; give them identical binary outcomes, so the always-survivor effect is zero. In model B one quarter are always survivors with $(Y(0),Y(1))=(0,1)$, one quarter survive only under zero with outcome $1$, one quarter survive only under one with outcome $0$, and one quarter never survive. Each observed arm again has survival probability one half and half its survivors at each outcome, but the always-survivor effect is one. Randomization cannot distinguish these complete observed laws. Mortality data remove missingness while leaving principal-stratum ambiguity.
The example also explains why a comparison of survivor means can equal zero while the always-survivor contrast is positive. Monotone survival, outcome bounds and further restrictions can narrow that target, but must be announced. They are distinct from the original unconditional cohort vector after assigning a mortality convention.
What this attempt establishes
Equations (1) and (2), the sharp missing-outcome bounds and the survivor counterexample give explicit checks for an eventual study. A convincing long-run analysis needs a fixed baseline cohort, documented regime variation, comparable adult measurement ages, tracing probabilities or missingness bounds, and stated survival outcomes. Age profiles alone do not identify sensitive periods; observed adult gains from a refugee-support policy do not identify the entire displacement contrast. No actual regime experiment, linked longitudinal data or transport evidence is supplied here. The catalogue's substantive vector of effects and its general persistence mechanisms therefore remain unresolved.Completion Estimate
45%This is a post hoc, heuristic estimate for the full catalogue target. The attempt derives explicit disruption-persistence stock paths, sharp missing-follow-up bounds, probability-tracing identification, and a survivor-effect nonidentification example. Full age-specific schooling, earnings, and health effects still require credible displacement-regime variation, comparable adult follow-up, family-separation mechanisms, and validated mortality and tracing assumptions.
References
- [refugees] S. Rozo and G. Grossman (2025), Refugees and Other Forcibly Displaced Populations, VoxDevLit 14(1), conclusions and future research. Primary review page: https://www.voxdev.org/voxdevlit/refugees-and-other-forcibly-displaced-populations/conclusions-future-research. The review supplies the research setting; the finite-horizon and identification examples above are self-contained illustrative models.