Economic interpretation of statistically identified shocks
LLM status Partially solved unverified
C32 — Time-Series Models - Dynamic Quantile Regressions - Dynamic Treatment Effect Models - Diffusion Processes - State Space Models Economics catalogue openaigpt 6 astra ultra 60% cataloguednot recordedPartially solvedStatement
ECON-020 | R | Source-backed agenda; exact formalization is editorial
Definitions and formal target
Suppose higher moments identify a structural innovation matrix up to a fully specified compatible set $\mathcal B(P)$, including any remaining rotations from observable law $P$. An economic shock label, such as a monetary or supply shock, must also satisfy substantively specified restrictions $\mathcal R$ involving external variables $Z$ or institutional timing.
Define the economically admissible response set at horizon $h$: \[ \Theta_h(P,\mathcal R)= \{e_y^{\mathsf T}A_h b_j:B\in\mathcal B(P),\ (B,j)\text{ satisfies }\mathcal R\}. \] Here $A_h$ is the reduced-form propagation matrix and $e_y$ selects the outcome. Determine when this set has a unique economically defensible interpretation and when ambiguity is irreducible with the available data.
Clarifications and required assumptions
For a stable reduced-form VAR $y_t=\sum_{\ell=1}^pa_\ell y_{t-\ell}+u_t$, the propagation matrices are $A_0=I$ and $A_h=\sum_{\ell=1}^{\min(p,h)}a_\ell A_{h-\ell}$. Column $b_j$ is the contemporaneous effect of shock $j$, and $e_y$ is a coordinate unit vector. Statistical identification may leave continuous rotational ambiguity as well as signs/permutations in degenerate cases; use the full compatible set $\mathcal B(P)$. Define the economic restriction set $\mathcal R$ before choosing a convenient shock label.
Variants and changes of scope
Unique target. The target is point-identified when $\Theta_h(P,\mathcal R)$ is a singleton; its nonemptiness is a separate compatibility check. Multiple horizons. Report a joint response-path set, preserving the same matrix and label across horizons. Proxy labeling. State the relevant covariances and exclusion assumptions; maximum sample correlation alone is a labeling convention, not causal identification. Interpretation. Narrative timing, sign, and zero restrictions have different economic content and can be inconsistent with each other.
Audit conclusion and source relationship
The exact source locator is Lewis, Section 6.4, on shock labeling. The response-set construction is editorial. Purely statistical independence does not supply a monetary, demand, or supply interpretation without additional substantive restrictions.
References and exact source roles
[S1] Daniel J. Lewis (2025). Identification Based on Higher Moments in Macroeconometrics. Annual Review of Economics 17, 665--693. DOI: 10.1146/annurev-economics-070124-051419. Checked published PDF in the author's institutional repository.
Verified locator / source role: Sections 6.4. These passages support the methodological agenda; the precise inferential criterion below is editorial.
https://discovery.ucl.ac.uk/id/eprint/10208055/1/Lewis_annurev-economics-070124-051419.pdf
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-508
Statistical shock identification and economic labeling
A finite labeling problem that can be solved exactly
Higher-moment identification and economic interpretation answer different questions. This attempt gives an exact algorithm when the statistical ambiguity consists of signed permutations, derives robust proxy-label conditions, and shows why a singleton response need not imply a unique economic label. Continuous rotational ambiguity and uncertain economic exclusions remain substantive limitations.Assume first that the population innovation matrix is known to belong to \[ \mathcal B=\{B_0PD:\ P\text{ a permutation matrix},\quad D=\operatorname{diag}(s_1,\ldots,s_m),\ s_j\in\{-1,1\}\}, \] where $B_0$ is nonsingular. This is an assumption about the full compatible set, not an automatic consequence of observing non-Gaussian residuals. For each candidate impact vector $b=\pm B_0e_j$, define its whole response path \[ r(b)=(e_y'A_0b,\ldots,e_y'A_Hb). \] Let $\mathcal R$ be specified predicates involving columns, observed external moments, and timing. If restrictions concern other labels as well, test the entire signed-permuted matrix, not just one column.
The distinction between a unique response and a unique label matters. Take $B_0=I_2$, and let the reported outcome be the sum of two observables, so its selector is $(1,1)'$. At horizon zero both positive columns have response one. If both are economically admissible, the response set is a singleton although the label is ambiguous. If a coordinate selector is required, add a third observable and use a nonsingular matrix with first row $(1,1,0)$; the first two columns still have the same first-coordinate impact. At later horizons the paths can separate. Thus identification of one scalar functional is weaker than identification of the shock's institutional meaning.
An external proxy gives testable labeling restrictions
Let $c=E[uZ]$ for a scalar external measurement $Z$, and calculate $d=B_0^{-1}c$. For candidate signed permutation $B=B_0PD$, \[ E[\varepsilon Z]=DP'd. \tag{1} \] An exact proxy exclusion for target index $j$ requires every nontarget coordinate of (1) to vanish and its target coordinate to be positive. Therefore, before labels are permuted, exactly one nonzero coordinate of $d$ is necessary and sufficient for a unique relevant column direction, with its sign fixed by positive relevance. If $d=0$, relevance fails; if two entries are nonzero, the exact exclusion is incompatible with every one-shock label in this model. This is more informative than assigning the largest correlation and silently discarding the others.Allow instead $|E[Z\varepsilon_k]|\le\epsilon$ for nontarget shocks and $E[Z\varepsilon_j]\ge\kappa>0$ for the target. A signed original column $sB_0e_j$ is admissible exactly when \[ sd_j\ge\kappa,\qquad \max_{k\ne j}|d_k|\le\epsilon. \tag{2} \] The signed-permutation assumptions make these inequalities sufficient as well as necessary: choose the matching permutation and signs, keeping the observed joint innovation-proxy law unchanged. This does not prove that the substantive bounded-exclusion assumption is true. It describes its implications within the declared statistical model.
For instance $d=(0.8,0.1,0)'$, $\epsilon=0.15$, $\kappa=0.5$ identifies the first positive column. The vector $(0.8,0.7,0)'$ with the same bounds makes every candidate fail, since either large coordinate would be a forbidden nontarget covariance. With $\epsilon=0.9$, $\kappa=0.5$, both first and second positive columns are admissible. These arithmetic examples illustrate rejection and ambiguity, not empirical proxy quality.
Uncertainty without a discontinuous winner rule
Suppose $B_0$ is known in a benchmark and an independent sample supplies $\widehat d$ with a simultaneous event $\|\widehat d-d\|_\infty\le r$ of probability at least $1-\alpha$. An outer confidence set of labels retains $(j,s)$ when \[ s\widehat d_j+r\ge\kappa,\qquad \max_{k\ne j} (|\widehat d_k|-r)_+\le\epsilon. \tag{3} \] Every truly admissible label satisfies (3) on that event, so projecting the retained labels to their paths covers the entire population path set with probability at least $1-\alpha$. The separate inequalities may admit a candidate without a jointly feasible nuisance vector when other restrictions are added; this makes the set conservative, never anti-conservative. With a coordinate box and just (2), feasibility can be checked directly coordinate by coordinate.When $B_0$ and the propagation matrices are estimated, their uncertainty must enter the same projection. A confidence region for $(B_0,A_0,\ldots,A_H,c)$ can be propagated through (1)--(3), excluding singular $B_0$ or allowing an uninformative result near singularity. Holding estimated impact directions fixed while varying only proxy correlations does not provide the claimed joint coverage. Covariance bounds based on concentration also require explicit moment or tail assumptions; arbitrary unbounded proxies do not satisfy a distribution-free Hoeffding bound.
A strict separation margin makes the labeling operation stable. If a true candidate satisfies its relevance inequality by more than $2r$ and all exclusion inequalities by more than $2r$, its empirical inequalities remain feasible. If every rival violates at least one inequality by more than $2r$, those rivals are rejected by (3). Without a positive margin, labels may remain multiple with arbitrarily large samples along local sequences. Reporting that ambiguity is the appropriate inferential behavior.
Continuous ambiguity cannot be ignored
Suppose two independent structural shocks are standard Gaussian and no higher-moment distinction separates them. Replacing their two columns by an arbitrary orthogonal rotation and rotating the shocks oppositely preserves their joint law. The compatible set then contains a circle, not merely signed permutations. Enumerating $2^m m!$ possibilities misses valid models. With $B_0=I_2$, the first coordinate's impact of a freely rotating unit column ranges over $[-1,1]$; a nonnegative-impact sign restriction reduces it to $[0,1]$, still a continuum.A valid nonzero one-shock proxy can select a direction in that block, but its exclusion has the same economic burden as before. A continuous compact rotation set can be projected numerically using nets and continuity bounds, with relaxed restrictions to avoid missing feasible boundary points. Finite output is then an approximation to the set, not a proof of point identification. The same rotation must be retained across all horizons; choosing a favorable column separately at each horizon creates impossible response paths.
Interpretation and unresolved content
The finite algorithm completely resolves labeling conditional on a correct finite compatible set and declared restrictions. It does not identify which external instrument actually excludes all other economic shocks, or whether measured information omits anticipated policy. Statistical independence can distinguish latent directions while leaving their institutional interpretation open. A successful application needs external measurements and timing restrictions strong enough to separate economic stories, as well as a test of their compatibility. No policy label or causal response is inferred from statistical independence alone.Completion Estimate
60%This is a post hoc, heuristic estimate for the full catalogue target. The attempt gives exact signed-permutation label enumeration, proxy compatibility and confidence restrictions, and examples separating unique responses from unique economic labels. Full interpretation still requires handling continuous compatible rotations and validating external exclusions and institutional timing rather than assuming a finite statistical ambiguity set.
References
- [lewis] D. J. Lewis (2025), Identification Based on Higher Moments in Macroeconometrics, Annual Review of Economics 17, 665--693, Section 6.4. https://discovery.ucl.ac.uk/id/eprint/10208055/1/Lewis_annurev-economics-070124-051419.pdf. The source discusses the shock-labeling issue; the finite procedures and illustrative moment vectors above state their own assumptions.