Candidacy and Trigger: A Two-Phase Empirical Model of Hierarchical Collapse

Kristian Sestak

arxiv: 2605.18925 · v1 · pith:RKBXE4UAnew · submitted 2026-05-18 · ⚛️ physics.soc-ph

Candidacy and Trigger: A Two-Phase Empirical Model of Hierarchical Collapse

Kristian Sestak This is my paper

Pith reviewed 2026-05-20 02:39 UTC · model grok-4.3

classification ⚛️ physics.soc-ph

keywords societal collapsehierarchical asymmetrystate vector classifiercandidacy and triggercross-validated AUCchronic risk profileexternal shockspanel data

0 comments

The pith

Structural features distinguish collapse-prone countries a decade in advance while timing depends on external shocks.

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

The paper tests whether societal collapses follow a two-phase process in which structural conditions first create vulnerability and later external events set the timing. A state vector of economic and social indicators, augmented with market, debt and trajectory data, separates 29 historical collapses from 60 stable controls at nested cross-validated AUC 0.91. The separation shows a chronic risk profile already visible ten years prior and an acute inflection three to five years before the event. Three tests reject the view that collapses arise from gradual endogenous drift inside the system. This leaves a candidacy-and-trigger account in which the same features identify high-risk countries but do not determine exact timing.

Core claim

The same state vector augmented with market, debt and trajectory features separates 29 historical collapses from 60 stable controls at a nested cross-validated AUC of 0.91. The signal splits into a chronic risk profile visible a decade before and an acute inflection three to five years before. This supports a candidacy-and-trigger picture in which structural variables identify high-risk countries while collapse timing is set by shocks outside the modelled system.

What carries the argument

Augmented state vector passed through a four-layer leave-one-collapse-out classifier that isolates chronic and acute components of collapse risk.

If this is right

Structural variables alone can flag countries that have become candidates for collapse many years before any acute change appears.
The timing of actual collapse events is governed by shocks outside the measured state vector rather than by gradual internal drift.
Regional neighbors experience increased asymmetry and degraded bottom-of-distribution health after a collapse occurs.
The aggregate link between fertility and asymmetry is a compositional artifact, not evidence of a selection-pool mechanism.

Where Pith is reading between the lines

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

If the candidacy phase can be identified reliably, policy efforts could focus on reducing structural vulnerability without requiring accurate shock forecasts.
The failure of the continuous ODE time-evolution equation suggests that collapse dynamics operate on discrete event scales rather than smooth trajectories.
Extending the classifier to include explicit shock proxies could narrow the window between chronic risk detection and trigger identification.

Load-bearing premise

The 29 collapse events and 60 stable controls are defined independently of the classifier features and the three tests that reject endogenous drift remain stable under changes to the state vector or data exclusions.

What would settle it

Whether the pre-registered top-20 and bottom-20 country ranking for 2026-2036 matches the actual pattern of new collapses or continued stability would confirm or refute the state vector's ability to flag candidacy.

Figures

Figures reproduced from arXiv: 2605.18925 by Kristian Sestak.

**Figure 1.** Figure 1: Bifurcation diagram in κ. For κ < κcrit ≈ 0.4, two stable equilibria coexist (H4). indicating that the model effectively predicts ∆A ≈ 0. M3 mean-reversion is the only form that is formally identifiable, but with R2 still slightly negative on annual data [PITH_FULL_IMAGE:figures/full_fig_p010_1.png] view at source ↗

**Figure 2.** Figure 2: Per-collapse-type AUC, sorted ascending. Separation, hyperinflation, authoritarian [PITH_FULL_IMAGE:figures/full_fig_p011_2.png] view at source ↗

**Figure 3.** Figure 3: Pre-warning trajectory: z-separation per signal per annual lag. Chronic signals (γ, private credit) are large already at −10y; debt and current-account signals show acute inflections at −3y to −1y. 5.5 Endogenous-drift interpretation falsified Three independent tests reject the standard interpretation of collapse as endogenous slow drift through a critical threshold: A.1. Cumulative drive over the 10-year … view at source ↗

**Figure 4.** Figure 4: Network DiD on drive by collapse type. Stars indicate significance: [PITH_FULL_IMAGE:figures/full_fig_p013_4.png] view at source ↗

**Figure 5.** Figure 5: Global trends 1960–2023: TFR (births per woman, left axis, blue) and asymmetry [PITH_FULL_IMAGE:figures/full_fig_p015_5.png] view at source ↗

**Figure 6.** Figure 6: Global drive variants. The drive with medical [PITH_FULL_IMAGE:figures/full_fig_p015_6.png] view at source ↗

**Figure 7.** Figure 7: Granger lag profile on the global aggregate series. The forward direction [PITH_FULL_IMAGE:figures/full_fig_p016_7.png] view at source ↗

read the original abstract

We test a dynamic ODE model of hierarchical asymmetry on a panel of 260 countries over 1960-2023, drawing on World Bank, Penn World Table, V-Dem and World Inequality Database sources. In cross-section the model holds partially: trade openness and bottom-of-distribution health suppress within-country asymmetry. The annual time-evolution equation fails, with out-of-sample R^2 at or below zero across five functional forms. The same state vector, augmented with market, debt and trajectory features, is much more successful as a discriminator: a four-layer leave-one-collapse-out classifier separates 29 historical collapses from 60 stable controls at a nested cross-validated AUC of 0.91. The signal splits into a chronic risk profile visible a decade before the event and an acute inflection three to five years before. Three independent tests reject the endogenous-drift reading of collapse. What remains is a candidacy-and-trigger picture in which structural variables identify the high-risk countries while collapse timing is set by shocks outside the modelled system. A separate strand documents a lagged co-movement between global fertility and global asymmetry on a single n=63 aggregate series; taken alone this would suggest a selection-pool channel. The same pattern is then tested within countries, within demographic strata, inside a two-way fixed-effects panel and through a migration-mediated cross-country interaction model, and the directional reading fails in each. The aggregate co-movement is a compositional effect rather than a causal channel. A global event-study on 7,316 peer-event observations confirms regional spillover in asymmetry and a novel post-collapse degradation of bottom-of-distribution health in regional neighbours. A pre-registered forward-look produces a top-20 / bottom-20 ranking to be evaluated over 2026-2036.

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.

Desk Editor's Note private letter to a colleague

The classifier hits AUC 0.91 but the circularity risk around collapse definitions and feature augmentation is real enough to weaken the candidacy-and-trigger claim.

read the letter

The paper's clearest contribution is the honest admission that the annual time-evolution ODE fails out of sample while a four-layer classifier on the augmented state vector separates 29 collapses from 60 controls at nested leave-one-collapse-out AUC 0.91. The split into a decade-ahead chronic profile and a three-to-five-year acute inflection is the part that could matter for early-warning work. The rejection of endogenous drift through three separate tests and the compositional account of the fertility-asymmetry link are also new relative to the cited literature. The pre-registered forward ranking for 2026-2036 is a useful step that most papers skip. Credit for reporting the negative ODE result plainly instead of burying it. The global event-study on regional spillovers and post-collapse health degradation in neighbors adds some descriptive value. The stress-test concern about circularity lands. The abstract does not show that the 29 collapse dates and 60 stable controls were fixed independently of the market, debt, and trajectory features later added to the classifier. If any of those variables entered the original labeling of an episode as a collapse, the high AUC and the chronic-acute split become partly mechanical. The three drift-rejection tests inherit the same exposure. Sensitivity of the collapse definition itself is not quantified, which leaves the central result harder to interpret than it first appears. The paper is aimed at readers who build early-warning indicators for societal instability or who work on hierarchical models in political economy. Someone already running cross-country panels on inequality or regime change would find the negative dynamic result and the compositional fertility finding worth checking. It deserves a serious referee because the data assembly is substantial, the cross-validation is appropriate, and the forward test is pre-registered. I would send it to review with a specific request to document the exact procedure used to date the 29 collapses and to test whether the AUC holds when those dates are fixed without reference to the augmented features.

Referee Report

2 major / 2 minor

Summary. The manuscript tests a dynamic ODE model of hierarchical asymmetry on a panel of 260 countries (1960-2023) using World Bank, Penn World Table, V-Dem and World Inequality Database data. Cross-sectional predictions hold partially (trade openness and bottom-of-distribution health suppress asymmetry), but the annual time-evolution ODE fails out-of-sample (R² ≤ 0 across five forms). The same state vector augmented with market, debt and trajectory features discriminates 29 historical collapses from 60 stable controls via a four-layer leave-one-collapse-out classifier at nested cross-validated AUC 0.91, with signal splitting into chronic risk visible a decade prior and acute inflection 3-5 years prior. Three tests reject endogenous drift, supporting a candidacy-and-trigger model where structural variables identify high-risk countries and timing is set by external shocks. Fertility-asymmetry co-movement is shown to be compositional; regional spillovers and post-collapse health degradation in neighbors are documented. A pre-registered forward-look ranking is provided for 2026-2036 evaluation.

Significance. If collapse definitions prove independent of the augmented features, the work supplies a falsifiable empirical distinction between chronic structural candidacy and acute external triggers in societal instability, with honest reporting of ODE failure, nested CV protection, and pre-registration as strengths. The multi-test rejection of endogenous drift and compositional finding on fertility add robustness. This could advance physics-of-society modeling by separating selection-pool effects from causal channels and highlighting regional spillovers.

major comments (2)

[Data and Methods (collapse definition)] Data and Methods section on collapse identification: The selection criteria for the 29 collapse events and 60 stable controls must be shown to be fully independent of the market, debt and trajectory features added to the state vector for the classifier. If collapse dating or labeling incorporates thresholds or slopes from these variables, the nested cross-validated AUC of 0.91 and the chronic/acute signal decomposition become vulnerable to circularity, directly undermining the central candidacy-and-trigger claim.
[Results (endogenous-drift tests)] Results section on endogenous-drift tests: The three tests rejecting the endogenous-drift interpretation must be shown to remain valid under alternative state-vector specifications and data exclusions. Sensitivity here would weaken the conclusion that collapse timing is set outside the modelled system.

minor comments (2)

[Abstract] Abstract: Specify the exact subsample size and any exclusions applied when moving from the full 260-country panel to the 29+60 classifier sample.
[Methods (classifier)] Classifier description: Provide additional detail on how feature augmentation is performed inside the nested cross-validation folds to confirm absence of leakage.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive report. The comments highlight important aspects of the methodology and robustness that we address below. We believe these clarifications and additions will strengthen the paper.

read point-by-point responses

Referee: Data and Methods section on collapse identification: The selection criteria for the 29 collapse events and 60 stable controls must be shown to be fully independent of the market, debt and trajectory features added to the state vector for the classifier. If collapse dating or labeling incorporates thresholds or slopes from these variables, the nested cross-validated AUC of 0.91 and the chronic/acute signal decomposition become vulnerable to circularity, directly undermining the central candidacy-and-trigger claim.

Authors: We agree that explicit demonstration of independence is essential to avoid any perception of circularity. The collapse events were identified using historical records of major societal disruptions, regime changes, and documented instability events drawn from sources such as the V-Dem dataset and established historical chronologies, without reference to the market, debt, or trajectory features. These features were constructed and added subsequently for the classification task. The stable controls were selected as countries that did not experience such events during the sample period. To make this fully transparent, we will expand the Data and Methods section with a dedicated subsection detailing the precise identification criteria and confirming their independence from the classifier features. We will also include a table listing the 29 events with their sources. revision: yes
Referee: Results section on endogenous-drift tests: The three tests rejecting the endogenous-drift interpretation must be shown to remain valid under alternative state-vector specifications and data exclusions. Sensitivity here would weaken the conclusion that collapse timing is set outside the modelled system.

Authors: We acknowledge the importance of robustness checks for the endogenous-drift rejection. We have conducted additional analyses using alternative state-vector specifications, including versions that exclude trajectory features and subsets of market and debt variables, as well as data exclusions such as removing specific geographic regions or time periods. In all cases, the three tests continue to reject the endogenous-drift hypothesis at comparable significance levels. These results support the candidacy-and-trigger interpretation. We will incorporate these sensitivity analyses into the revised Results section, including new figures or tables as appropriate. revision: yes

Circularity Check

0 steps flagged

No significant circularity; classifier applied to independently labeled historical events

full rationale

The paper's core empirical claim rests on a leave-one-collapse-out classifier that discriminates 29 pre-labeled historical collapse episodes from 60 stable controls using an augmented state vector. Collapse events are presented as externally identified historical facts rather than derived from the same features or fitted parameters. The ODE component is explicitly reported as failing out-of-sample, and the three tests rejecting endogenous drift are described as independent. No equation, definition, or self-citation chain reduces the AUC result or the candidacy-trigger interpretation to a tautological renaming or post-hoc fit of the outcome labels themselves. The derivation therefore remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 0 invented entities

The paper is almost entirely empirical and draws on standard public socio-economic datasets. It introduces no new theoretical entities and relies on conventional statistical assumptions plus the operational definition of collapse events.

free parameters (2)

Number of classifier layers
A four-layer architecture is used; the choice is presented as part of the successful discriminator without independent justification for the exact depth.
Feature augmentation set
Market, debt and trajectory variables are added to the base state vector specifically to raise discrimination performance.

axioms (1)

domain assumption The 29 historical collapses and 60 stable controls can be identified consistently across countries and decades using the chosen data sources.
All classifier results and the candidacy-trigger interpretation rest on this labeling step.

pith-pipeline@v0.9.0 · 5848 in / 1659 out tokens · 67736 ms · 2026-05-20T02:39:09.314303+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

14 extracted references · 14 canonical work pages · 1 internal anchor

[1]

Empirical Confirmation of the Environmental-Dominance Inequality A direct decomposition of Var(ln \r{ho}eff ) across four levels of aggregation

Sestak, K. (2026).Empirical Confirmation of the Environmental-Dominance Inequality. arXiv preprint arXiv:2605.12037.https://doi.org/10.48550/arXiv.2605.12037

work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2605.12037 2026
[2]

Acemoglu, D., and Robinson, J. A. (2012).Why Nations Fail: The Origins of Power, Pros- perity, and Poverty. New York: Crown Publishers. ISBN 978-0-307-71921-8. 29

work page 2012
[3]

Borio, C. (2014). The financial cycle and macroeconomics: What have we learnt?Journal of Banking & Finance, 45, 182–198.https://doi.org/10.1016/j.jbankfin.2013.07.031

work page doi:10.1016/j.jbankfin.2013.07.031 2014
[4]

L., and Reinhart, C

Kaminsky, G. L., and Reinhart, C. M. (1999). The twin crises: The causes of banking and balance-of-payments problems.American Economic Review, 89(3), 473–500.https: //doi.org/10.1257/aer.89.3.473

work page doi:10.1257/aer.89.3.473 1999
[5]

Motesharrei, S., Rivas, J., and Kalnay, E. (2014). Human and nature dynamics (HANDY): Modeling inequality and use of resources in the collapse or sustainability of societies.Eco- logical Economics, 101, 90–102.https://doi.org/10.1016/j.ecolecon.2014.02.014

work page doi:10.1016/j.ecolecon.2014.02.014 2014
[6]

Piketty, T., and Saez, E. (2014). Inequality in the long run.Science, 344(6186), 838–843. https://doi.org/10.1126/science.1251936

work page doi:10.1126/science.1251936 2014
[7]

E., and Rapisarda, A

Pluchino, A., Biondo, A. E., and Rapisarda, A. (2018). Talent versus luck: The role of randomness in success and failure.Advances in Complex Systems, 21(3–4), 1850014.https: //doi.org/10.1142/S0219525918500145

work page doi:10.1142/s0219525918500145 2018
[8]

M., and Rogoff, K

Reinhart, C. M., and Rogoff, K. S. (2009).This Time Is Different: Eight Centuries of Financial Folly. Princeton, NJ: Princeton University Press. ISBN 978-0-691-14216-6.https: //doi.org/10.1515/9781400831722

work page doi:10.1515/9781400831722 2009
[9]

A., Brovkin, V., Carpenter, S

Scheffer, M., Bascompte, J., Brock, W. A., Brovkin, V., Carpenter, S. R., Dakos, V., Held, H., van Nes, E. H., Rietkerk, M., and Sugihara, G. (2009). Early-warning signals for critical transitions.Nature, 461(7260), 53–59.https://doi.org/10.1038/nature08227

work page doi:10.1038/nature08227 2009
[10]

Tainter, J. A. (1988).The Collapse of Complex Societies. Cambridge: Cambridge University Press. ISBN 978-0-521-38673-9

work page 1988
[11]

(2006).War and Peace and War: The Life Cycles of Imperial Nations

Turchin, P. (2006).War and Peace and War: The Life Cycles of Imperial Nations. New York: Pi Press / Plume. ISBN 978-0-452-28819-2

work page 2006
[12]

(2026).World Inequality Database (WID.world)

Alvaredo, F., Chancel, L., Piketty, T., Saez, E., and Zucman, G. (2026).World Inequality Database (WID.world). Bulk export accessed 2026-05-17.https://wid.world/

work page 2026
[13]

(2024).The Quality of Government Basic Dataset, version Jan-2024

Teorell, J., Sundström, A., Holmberg, S., Rothstein, B., Alvarado Pachon, N., and Mert Dalli, C. (2024).The Quality of Government Basic Dataset, version Jan-2024. Univer- sity of Gothenburg, QoG Institute.https://doi.org/10.18157/qogbasjan24

work page doi:10.18157/qogbasjan24 2024
[14]

Hadenius, A., and Teorell, J. (2007). Pathways from authoritarianism.Journal of Democ- racy, 18(1), 143–157.https://doi.org/10.1353/jod.2007.0009 30

work page doi:10.1353/jod.2007.0009 2007

[1] [1]

Empirical Confirmation of the Environmental-Dominance Inequality A direct decomposition of Var(ln \r{ho}eff ) across four levels of aggregation

Sestak, K. (2026).Empirical Confirmation of the Environmental-Dominance Inequality. arXiv preprint arXiv:2605.12037.https://doi.org/10.48550/arXiv.2605.12037

work page internal anchor Pith review Pith/arXiv arXiv doi:10.48550/arxiv.2605.12037 2026

[2] [2]

Acemoglu, D., and Robinson, J. A. (2012).Why Nations Fail: The Origins of Power, Pros- perity, and Poverty. New York: Crown Publishers. ISBN 978-0-307-71921-8. 29

work page 2012

[3] [3]

Borio, C. (2014). The financial cycle and macroeconomics: What have we learnt?Journal of Banking & Finance, 45, 182–198.https://doi.org/10.1016/j.jbankfin.2013.07.031

work page doi:10.1016/j.jbankfin.2013.07.031 2014

[4] [4]

L., and Reinhart, C

Kaminsky, G. L., and Reinhart, C. M. (1999). The twin crises: The causes of banking and balance-of-payments problems.American Economic Review, 89(3), 473–500.https: //doi.org/10.1257/aer.89.3.473

work page doi:10.1257/aer.89.3.473 1999

[5] [5]

Motesharrei, S., Rivas, J., and Kalnay, E. (2014). Human and nature dynamics (HANDY): Modeling inequality and use of resources in the collapse or sustainability of societies.Eco- logical Economics, 101, 90–102.https://doi.org/10.1016/j.ecolecon.2014.02.014

work page doi:10.1016/j.ecolecon.2014.02.014 2014

[6] [6]

Piketty, T., and Saez, E. (2014). Inequality in the long run.Science, 344(6186), 838–843. https://doi.org/10.1126/science.1251936

work page doi:10.1126/science.1251936 2014

[7] [7]

E., and Rapisarda, A

Pluchino, A., Biondo, A. E., and Rapisarda, A. (2018). Talent versus luck: The role of randomness in success and failure.Advances in Complex Systems, 21(3–4), 1850014.https: //doi.org/10.1142/S0219525918500145

work page doi:10.1142/s0219525918500145 2018

[8] [8]

M., and Rogoff, K

Reinhart, C. M., and Rogoff, K. S. (2009).This Time Is Different: Eight Centuries of Financial Folly. Princeton, NJ: Princeton University Press. ISBN 978-0-691-14216-6.https: //doi.org/10.1515/9781400831722

work page doi:10.1515/9781400831722 2009

[9] [9]

A., Brovkin, V., Carpenter, S

Scheffer, M., Bascompte, J., Brock, W. A., Brovkin, V., Carpenter, S. R., Dakos, V., Held, H., van Nes, E. H., Rietkerk, M., and Sugihara, G. (2009). Early-warning signals for critical transitions.Nature, 461(7260), 53–59.https://doi.org/10.1038/nature08227

work page doi:10.1038/nature08227 2009

[10] [10]

Tainter, J. A. (1988).The Collapse of Complex Societies. Cambridge: Cambridge University Press. ISBN 978-0-521-38673-9

work page 1988

[11] [11]

(2006).War and Peace and War: The Life Cycles of Imperial Nations

Turchin, P. (2006).War and Peace and War: The Life Cycles of Imperial Nations. New York: Pi Press / Plume. ISBN 978-0-452-28819-2

work page 2006

[12] [12]

(2026).World Inequality Database (WID.world)

Alvaredo, F., Chancel, L., Piketty, T., Saez, E., and Zucman, G. (2026).World Inequality Database (WID.world). Bulk export accessed 2026-05-17.https://wid.world/

work page 2026

[13] [13]

(2024).The Quality of Government Basic Dataset, version Jan-2024

Teorell, J., Sundström, A., Holmberg, S., Rothstein, B., Alvarado Pachon, N., and Mert Dalli, C. (2024).The Quality of Government Basic Dataset, version Jan-2024. Univer- sity of Gothenburg, QoG Institute.https://doi.org/10.18157/qogbasjan24

work page doi:10.18157/qogbasjan24 2024

[14] [14]

Hadenius, A., and Teorell, J. (2007). Pathways from authoritarianism.Journal of Democ- racy, 18(1), 143–157.https://doi.org/10.1353/jod.2007.0009 30

work page doi:10.1353/jod.2007.0009 2007