pith. sign in

arxiv: 2512.17005 · v2 · submitted 2025-12-18 · 💰 econ.EM · stat.ME

Principled Identification of Structural Dynamic Models

Neville Francis , Peter Reinhard Hansen , Chen Tong This is my paper

Pith reviewed 2026-05-16 21:10 UTC · model grok-4.3

classification 💰 econ.EM stat.ME
keywords SVAR identificationstructural shocksorthogonal rotationcorrelation maximizationproxy VAROASISCholesky orderingdynamic models
0
0 comments X

The pith

A weighted correlation-maximizing objective produces an identification scheme for structural shocks that aligns more closely with target variables than recursive methods.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper reframes identification in structural dynamic models as an optimization task that selects the orthogonal rotation maximizing a weighted sum of correlations with chosen target variables. This yields the Order- and Scale-Invariant Scheme, called OASIS, which the authors position as a natural reference rotation when the targets are the reduced-form innovations themselves. If the approach holds, recursive Cholesky identifications become a constrained special case of the same objective, and OASIS closes roughly twice the remaining gap to perfect correlation, both in theoretical comparisons and in re-examinations of published SVAR applications. The framework further supplies a symmetric way to handle multiple instruments and leakage in proxy VARs, while empirical checks on 22 existing studies show that reduced-form innovations are typically only weakly correlated.

Core claim

Rather than imposing restrictions alone, identification proceeds by optimizing a weighted correlation-maximizing objective that selects the orthogonal rotation most aligned with designated target variables. This produces the Order- and Scale-Invariant Scheme (OASIS). Recursive Cholesky identification is recovered as a constrained version of the identical objective. OASIS is systematically closer to perfect correlation, closing roughly twice as much of the gap as recursive orderings both theoretically and empirically. The same objective supplies a principled estimation strategy for multi-proxy IV-SVARs that resolves overdetermination while symmetrically accommodating proxy leakage.

What carries the argument

OASIS, the Order- and Scale-Invariant Scheme obtained by maximizing a weighted sum of correlations between rotated shocks and designated target variables.

If this is right

  • Recursive Cholesky orderings emerge directly as a special case of the OASIS objective under particular constraints on the weights and targets.
  • OASIS closes roughly twice the gap to perfect correlation with the targets relative to any recursive ordering, in both analytic bounds and numerical illustrations.
  • The weighted objective resolves overdetermination when multiple proxies are available and treats leakage from each proxy symmetrically.
  • In 22 published SVAR applications the reduced-form innovations exhibit only weak average correlations, which accounts for the historical robustness of recursive schemes.
  • Reapplying OASIS to existing proxy-VAR studies uncovers material cross-shock leakage that can change the substantive conclusions drawn from the same data.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The optimization perspective could be ported to identification problems in other classes of linear dynamic models where orthogonal rotations must be chosen.
  • Applied researchers could treat OASIS correlations as a quantitative benchmark to gauge how much their chosen economic restrictions improve alignment beyond a purely statistical baseline.
  • Accounting for leakage in previously published proxy-VAR results may revise the magnitude or even the sign of estimated impulse responses in several well-known applications.

Load-bearing premise

That choosing target variables and maximizing their weighted correlations with the rotated shocks recovers the economically relevant structural identification, which still requires exogenous economic insight to select the targets.

What would settle it

Simulate data from a known dynamic model with true orthogonal shocks, apply both OASIS and recursive orderings to the reduced-form residuals, and test whether the OASIS correlations with the designated targets exceed those of the recursive scheme by approximately the factor claimed.

Figures

Figures reproduced from arXiv: 2512.17005 by Chen Tong, Neville Francis, Peter Reinhard Hansen.

Figure 1
Figure 1. Figure 1: is an illustration of some of the results in [PITH_FULL_IMAGE:figures/full_fig_p014_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Scatterplot of − log(1 − ρ¯) against log d(C). The values shown along the axes are ¯ρ and d(C). OASIS points lie near the line y = −x+log 8, whereas Cholesky points lie near y = −x+log 4. observed that the values of ¯ρc do not systematically lie above or below the dashed line for Cholesky, so it does not appear that the orderings in the empirical literature were selected with this in mind. By construction,… view at source ↗
read the original abstract

We take a new perspective on identification in structural dynamic models: rather than imposing restrictions alone, we optimize an objective. While definitive structural identification ultimately requires exogenous economic insight, a weighted correlation-maximizing objective yields an Order- and Scale-Invariant Scheme (OASIS) that selects the orthogonal rotation most aligned with designated target variables. In traditional SVARs, these targets are the reduced-form innovations, making OASIS a natural reference rotation. We show that recursive Cholesky identification is a constrained version of the same objective and that OASIS is systematically closer to perfect correlation, closing roughly twice as much of the gap as recursive orderings, both theoretically and empirically. The same framework also provides a principled estimation strategy for Proxy VARs (IV-SVARs), where the weighted criterion is essential for resolving overdetermination in multi-proxy systems while symmetrically accommodating proxy leakage. Revisiting 22 published SVARs, we find that reduced-form innovations are typically only weakly correlated, helping explain the historical robustness of recursive schemes. Applying OASIS to seminal proxy applications, however, reveals economically important leakage across shocks and shows that accounting for such leakage can materially alter substantive conclusions.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The manuscript proposes an Order- and Scale-Invariant Scheme (OASIS) for identification in structural dynamic models. Rather than relying solely on restrictions, it optimizes a weighted correlation-maximizing objective over orthogonal rotations to select the rotation most aligned with designated target variables (reduced-form innovations in the SVAR case). Recursive Cholesky is shown to be a constrained special case of the same objective. The paper claims OASIS is systematically closer to perfect correlation than recursive orderings, closing roughly twice as much of the gap both theoretically and in 22 SVAR applications. It further extends the framework to Proxy VARs to resolve overdetermination in multi-proxy systems while symmetrically handling proxy leakage, with applications revealing economically important leakage that alters substantive conclusions.

Significance. If the central claims hold, OASIS provides a principled, objective-driven reference rotation for SVAR and Proxy VAR identification that improves upon ad-hoc recursive schemes while accommodating proxy leakage. The empirical finding that reduced-form innovations are typically only weakly correlated across 22 published SVARs offers a useful explanation for the historical robustness of recursive methods and could inform future identification choices in macroeconometrics.

major comments (2)
  1. [Abstract / Empirical applications] The quantitative claim that OASIS closes roughly twice as much of the gap to perfect correlation as recursive orderings (both theoretically and empirically) is load-bearing for the 'systematically' assertion. This factor-of-two advantage necessarily depends on the specific target weights and designations used in the weighted objective; no sensitivity checks to uniform weights, alternative target subsets, or small perturbations around the reported weights are described.
  2. [Proxy VAR extension] The abstract states that the weighted criterion is essential for resolving overdetermination in multi-proxy systems, but the precise mechanism by which the objective achieves this without introducing additional bias or requiring further restrictions is not derivable from the provided summary. A concrete illustration with the multi-proxy case (including how weights are chosen) is needed to substantiate the claim.
minor comments (2)
  1. [Abstract] The abstract would benefit from an explicit statement of the functional form of the weighted correlation objective (e.g., the precise definition of the weights and correlation measure) to allow readers to assess invariance properties immediately.
  2. [Introduction / Method] Notation for 'target variables' and 'designated targets' should be introduced with a short formal definition in the main text to avoid ambiguity when the same framework is applied to both SVAR innovations and external proxies.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive and insightful comments. We address each major comment below and outline revisions that will strengthen the manuscript without altering its core claims.

read point-by-point responses

  1. Referee: [Abstract / Empirical applications] The quantitative claim that OASIS closes roughly twice as much of the gap to perfect correlation as recursive orderings (both theoretically and empirically) is load-bearing for the 'systematically' assertion. This factor-of-two advantage necessarily depends on the specific target weights and designations used in the weighted objective; no sensitivity checks to uniform weights, alternative target subsets, or small perturbations around the reported weights are described.

    Authors: We agree that the factor-of-two advantage is central to the 'systematically closer' claim and that its robustness to weight choices merits explicit verification. In the revised manuscript we will add a dedicated sensitivity section that reports results under (i) uniform weights, (ii) alternative target subsets that drop or swap one variable at a time, and (iii) small additive perturbations (±10 percent) to the baseline weights. These checks preserve the qualitative ordering that OASIS closes substantially more of the gap than any recursive scheme, although the precise numerical factor varies modestly with the weighting scheme. revision: yes

  2. Referee: [Proxy VAR extension] The abstract states that the weighted criterion is essential for resolving overdetermination in multi-proxy systems, but the precise mechanism by which the objective achieves this without introducing additional bias or requiring further restrictions is not derivable from the provided summary. A concrete illustration with the multi-proxy case (including how weights are chosen) is needed to substantiate the claim.

    Authors: We appreciate the request for a concrete illustration. The weighted objective resolves overdetermination by treating the problem as a continuous optimization over the orthogonal group: each proxy receives a weight proportional to its first-stage strength (or an economic prior), and the optimizer selects the unique rotation that maximizes the weighted sum of squared correlations. Because the objective is strictly concave in the relevant subspace when weights are positive and distinct, the solution is unique without further restrictions and does not introduce bias beyond the maintained assumption that the proxies are valid instruments for their target shocks. In the revision we will insert a short numerical example (a three-variable, two-proxy simulation) that shows the weight vector, the resulting rotation matrix, and the contrast with an unweighted or equally weighted alternative. revision: yes

Circularity Check

1 steps flagged

Theoretical claim that OASIS is closer to perfect correlation reduces to the definition of its objective function

specific steps
  1. self definitional [Abstract]
    "a weighted correlation-maximizing objective yields an Order- and Scale-Invariant Scheme (OASIS) that selects the orthogonal rotation most aligned with designated target variables. ... We show that recursive Cholesky identification is a constrained version of the same objective and that OASIS is systematically closer to perfect correlation, closing roughly twice as much of the gap as recursive orderings, both theoretically and empirically."

    OASIS is explicitly defined as the rotation maximizing the weighted correlation with targets. Cholesky is stated to be a constrained version of the identical objective. Therefore the claim that OASIS is 'closer to perfect correlation' (higher correlation value) holds by construction as the maximizer, not as a separate theoretical derivation.

full rationale

The paper defines OASIS via a newly proposed weighted correlation-maximizing objective over orthogonal rotations. It then states that OASIS is systematically closer to perfect correlation than recursive Cholesky (a constrained special case of the same objective), both theoretically and empirically. The theoretical component follows directly by construction from the maximization, rendering the 'closer' result tautological rather than independently derived. The empirical summary in 22 applications and the factor-of-two gap closure are not circular but depend on the specific weights and targets chosen. No self-citation chains, imported uniqueness theorems, or ansatz smuggling are evident from the provided text. This yields partial circularity focused on the definitional theoretical claim.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The central claim rests on the standard SVAR assumption that structural shocks are orthogonal and that the reduced-form covariance matrix admits an orthogonal rotation to structural form. The new element is the choice of objective function and target variables, which are not derived from first principles but proposed as a modeling choice.

free parameters (1)
  • target weights
    The weighted correlation objective requires choosing or estimating weights for the designated target variables.
axioms (2)
  • domain assumption Structural shocks are mutually orthogonal
    Standard assumption in SVAR literature that allows decomposition of the reduced-form covariance matrix up to rotation.
  • standard math Reduced-form innovations can be rotated via an orthogonal matrix to recover structural shocks
    Follows from the Cholesky or spectral decomposition of the covariance matrix.

pith-pipeline@v0.9.0 · 5497 in / 1373 out tokens · 49328 ms · 2026-05-16T21:10:29.929438+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

3 extracted references · 3 canonical work pages

  1. [1]

    and Hansen, P

    Archakov, I. and Hansen, P. R. (2024). A canonical representation of block matrices with applications to covariance and correlation matrices.Review of Economics and Statistics, 106:1099–1113. Basu, S. and Bundick, B. (2017). Uncertainty shocks in a model of effective demand.Econo- metrica, 85:937–958. Baumeister, C. and Hamilton, J. D. (2015). Sign restri...

    work page 2024

  2. [2]

    Engle, R. F. and Issler, J. V. (1995). Estimating common sectoral cycles.Journal of Monetary Economics, 35:83–113. Faust, J. and Leeper, E. M. (1997). When do long-run identifying restrictions give reliable results?Journal of Business and Economic Statistics, 15:345–353. Forni, M. and Gambetti, L. (2016). Government spending shocks in open economy vars. J...

    work page arXiv 1995

  3. [3]

    same” if replication codes were available and the original sample period was used for estimation. Otherwise we give our estimation sample period

    =E ij +O(E 2), such that summing overiin (9) yields tr(L) =n− 1 2 nX i=1 X j<i E2 ij +O tr{E3} . Sinced(C) = 2 n P i>j E2 ij, we can conclude that 1 n tr( ˜A′ cC) = 1− 1 4 d(C) +O tr{E3} . Combined, we have 1−¯ρc 1−¯ρ∗ = 21 +O(tr{E 3}/d(C)) 1 +O(tr{E 3}/d(C)) = 2 (1 +O(∥E∥ F )) = 2 1 +O p d(C) , using tr{E 3}=O(∥E∥ 3 F ) andd(C) =∥E∥ 2 F /n=O(∥E∥ 2 F ).□ ...

    work page 2024