REVIEW 3 major objections 5 minor 1 cited by
Bank Runs With and Without Bank Failure
T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash
Pith's one-line read Runs rarely kill healthy banks: 3,421 U.S. bank runs from 1863 to 1934 show failure concentrates in weak institutions.
desk verdict A genuinely new bank-run dataset that supports the fundamentals-matter-more-than-self-fulfilling-panics view, but the paper's strongest causal claim rests on a solvency proxy that the authors admit is incomplete. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The central object is a new bank-distress-episodes database built by applying large language models to historical newspaper scans, classifying each episode into run only, run-suspension-reopening, or run with failure, and matching banks to annual national-bank balance sheets. The argument's load-bearing measure is a fundamentals index: the negative of a recursively estimated predicted failure probability from a regression of next-year receivership on balance-sheet ratios such as surplus-to-equity, noncore funding, liquid assets, deposits-to-assets, and asset growth. This index separates banks into weak and strong terciles and deciles; the paper then estimates pass-through regressions of failure on runs interacted with the index, local projections of deposits and loans, and city-level impulse responses for manufacturing. A secondary mechanism is textual classification of bank responses, including accommodation of withdrawals, equity injection, borrowing, partial or full suspension, and clearinghouse examination, which explains why strong banks survive.
What would settle it
Find a set of banks in the top decile of the fundamentals index that failed after a run and whose receivership records show high recovery rates and examiner-assessed asset quality, meaning they were genuinely solvent when they failed. If such cases are numerous, well above the estimated 4 percent failure probability for that decile, the necessity claim fails; if examiner and recovery data confirm that top-decile failure cases were actually insolvent, the run-induced-failure channel for healthy banks is essentially empty.
Extended reading notes
Core claim
On its own terms, the paper establishes that bank runs in the pre-FDIC United States were common but rarely fatal to sound institutions. Using text classifications of contemporary newspapers, it documents 3,421 runs, about 1,515 of which ended in failure. Conditional on a run, banks in the lowest decile of a fundamentals index fail with probability around 63 percent; for banks in the top decile the failure probability is estimated at about 4 percent and statistically indistinguishable from zero. Runs that newspapers identify as stemming from misinformation or confusion (44 national-bank cases) carry an 11 percent failure probability, and failures after such runs occur only in weak or fraudulent banks. At the city level, runs on weak banks are followed by declines of roughly 40 to 60 percent in local deposits and loans and a 5 percent fall in manufacturing activity, while runs on strong banks and non-fundamental runs show small or no effects. The paper reads this as evidence that poor fundamentals, not liquidity panics themselves, are what turn runs into failures and into real economic damage.
Load-bearing premise
The load-bearing assumption is that annual balance-sheet data reveal which banks were truly weak before a run; if many banks classified as strong were secretly insolvent or hid losses, the claim that strong banks almost never fail in runs would be overstated.
Editorial extensions
If this is right
- If the central claim holds, bank runs should be modeled as symptoms or amplifiers of underlying insolvency rather than as exogenous shocks that independently determine bank failure.
- Policy attention should focus on detecting and repairing weak bank fundamentals; pure liquidity support is not enough to prevent failure when fundamentals are poor, but suspension and interbank assistance can protect sound banks.
- Empirical studies of banking crises that measure distress only through bank failures will miss most runs and will overstate the relationship between runs and failure.
- Expect runs without failure to be substantially more common during panics, and local deposit and loan contractions to be concentrated where weak banks are run.
- The 11 percent failure probability of non-fundamental runs gives a concrete upper bound on how often pure panic can kill a bank in this historical environment.
Reading between the lines
- An implicit extension is that the 1863-1934 U.S. institutional environment, with no deposit insurance, limited lender of last resort, and branching restrictions, may make runs more frequent than today, but the fundamental-contingent failure pattern could persist in modern runs; testing it on recent uninsured-deposit runs would be a natural extension.
- The paper's reliance on observable annual balance sheets means hidden insolvency such as fraud or unrecognized losses could move some 'strong-bank' failures into the fundamentals camp; linking run outcomes to ex-post recovery rates and examiner asset-quality ratings would test this.
- The non-fundamental-run sample is small, only 44 national-bank cases, so the 11 percent estimate is imprecise; a larger corpus or cross-country historical newspapers could sharpen it.
- City-level reallocation effects, where deposits flee a run bank to other local banks, may explain why non-fundamental runs have no local effect; bank-level and city-level results together suggest redistribution rather than destruction.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper constructs a new dataset of 3,421 bank runs in U.S. newspapers from 1863 to 1934 using large language models, links these runs to national-bank balance sheets and official OCC receivership records, and studies the determinants and consequences of runs. The authors find that runs occur disproportionately in weak banks but also occur in strong banks, especially after negative aggregate or local news; that conditional on a run, failure is far more likely in weak banks, with a top-decile failure probability of about 4 percent that is not statistically significant; that strong banks typically survive runs through suspension, signaling, equity injections, and interbank support; and that local declines in deposits, lending, and manufacturing are concentrated in runs on weak banks or in runs with failure. The paper interprets these patterns as evidence that poor fundamentals are necessary for runs to translate into failure and severe real effects, tempering the view that small shocks can trigger costly self-fulfilling panics.
Significance. If the results hold, this is a major empirical contribution. The new run database is validated against official OCC failure records, narrative crisis chronologies, and an independent human audit with a 95 percent match rate; the authors are transparent about underreporting of small-bank runs and publicly document the episodes. The finding that most runs do not result in failure, and that failure pass-through is strongly graded in observable fundamentals, directly informs the debate between fundamentals-based and multiple-equilibrium views of banking panics. The paper also provides novel descriptive evidence on how strong banks survive runs and on the local real effects of runs with and without failure. The main risks are that the headline 'necessity' claim is stronger than the statistical tests support, and that the observational local projections cannot fully rule out confounding by local economic trends.
major comments (3)
- [Section 6.1, Figure 5, Table 5] The central claim that weak fundamentals are necessary for a run to be associated with failure is identified off the annual call-report fundamentals index. The paper itself states in Section 6.1 that 'publicly available financial statements cannot capture episodes where a bank was insolvent but had not yet recognized losses'; this is exactly the alternative reading of the 4 percent top-decile failure probability: those runs may have hit banks that were already insolvent on an economic basis but looked strong on observable book capital. The large-bank robustness (Table 5, columns 5-6) addresses underreporting of runs, not mismeasurement of solvency, and the Appendix A.1 event studies show failing banks have weak observables on average but do not examine the few top-decile banks that fail after runs. Please use OCC examiner assessments, as in Correia, Luck and Verner (2025), for the strong-fundamentals banks that fail following runs, or otherwise bound the extent of hidden insolvency, and state explicitly how the 'necessary' conclusion is affected.
- [Section 6.1, Figure 5, and Section 8] The claim that weak bank fundamentals are 'necessary' for runs to cause failure is not supported by the reported test. The top-decile conditional failure probability is 4 percent and statistically insignificant; failing to reject zero is not evidence that the probability is exactly zero. The abstract and conclusion state necessity, but the evidence can at most support 'runs on the healthiest banks are rarely followed by failure.' Please report the confidence interval for the top-decile estimate, use language consistent with the statistical precision, or conduct an explicit equivalence or bounding exercise that justifies the necessity wording.
- [Section 7.2, Equations (11)-(16), Figures 9-10] The local projections that distinguish runs on weak versus strong banks are observational, and the paper's own Table 7 shows that runs are more likely after local business failures, so cities experiencing weak-bank runs may be on differentially declining trends. The non-fundamental-run design (Section 6.3, based on 44 runs) is a useful step, but the city-level non-fundamental-run indicators are extremely sparse, so the claim that pure liquidity runs have no significant local effects may be underpowered. Please add a formal discussion of pre-trends, alternative control groups, or sensitivity bounds, and temper the causal language in Section 8 accordingly.
minor comments (5)
- [Abstract (arXiv metadata)] The abstract in the article header reports 3,984 runs, while the full-text abstract and Section 4 report 3,421 runs; please reconcile the two numbers.
- [Table 1 and Figure 1] The suspension total in Table 1 Panel A (13,358) does not reconcile with the episode-type counts implied by Figure 1 (run only 1,325, suspension only 2,158, run-suspension-reopening 581, suspension-failure 8,815, run-suspension-failure 1,515, which sum to 13,069 suspensions). Please check the counts or clarify the discrepancy.
- [Tables 3 and 5] The units of the dependent variable appear inconsistent across tables: Table 3 reports a mean dependent variable of 0.33 (consistent with a rate expressed in percent), while Table 5 reports 0.0085 (a decimal proportion). Please make the units uniform or label them clearly in the table notes.
- [Section 7.2.2] The sentence describing the manufacturing index says 'every month from 1933 to 1935' and later 'monthly for the remainder of 1933 through 1933'; the latter appears to be a typo for 1935. Please also clarify how the forward fill from monthly to weekly frequencies is implemented.
- [Section 5.3] The AUC values in Table 4 are described as in-sample. Please note explicitly in the text or table notes that these are in-sample fits, since out-of-sample predictive power may be lower, particularly for the run-without-failure panel where the AUC is close to 0.6 in column 1.
Circularity Check
Minor overlap: the fundamentals index is fitted to failure outcomes and then used to describe failure after runs, but the central run-interaction result is not forced by that fit.
-
fitted input called prediction
[Section 6.1, Equation (3), Table 5, and Figure 5]
"We start by constructing a simple bank-level measure of fundamentals, denoted Fundamentalsbt, based on the regression of bank failure on bank fundamentals ... Fundamentalsbt is then defined as the negative of the predicted value from the regression that predicts failure in year t+1 based on bank fundamentals. ... For banks in the strongest decile of our proxy for fundamentals, the probability of failure during a run falls to 4% and is not statistically significant. Thus, we cannot reject that runs on the healthiest banks never cause bank failures."
The Fundamentals index is constructed as the negative of the predicted failure probability from an in-sample regression of bank failure on balance-sheet variables. Sorting banks into deciles of this index mechanically ranks them by fitted failure risk, so the very low unconditional failure rate in the top decile is built into the index. The claim that runs on 'strong' banks rarely end in failure is therefore partly a re-description of the first-stage fit rather than an independent out-of-sample prediction. However, the pass-through result that matters for the paper's central claim is the interaction of the run indicator with Fundamentals; that interaction is not estimated in the first stage and could in principle have been zero.
full rationale
The paper's main derivation chain is otherwise self-contained. The novel bank-run database is validated against external narrative chronologies (Jalil 2015; Baron, Verner and Xiong 2021) and against OCC receivership records and deposit-outflow data, which are independent benchmarks. The theoretical predictions in Section 2 come from standard models in the literature, not from the paper's own estimates. The central new empirical content is the differential pass-through of runs to failure and to local outcomes across banks of different health. This differential effect is identified from the interaction of the run dummy with the fundamentals measure; the first-stage failure regression used to build the fundamentals index does not include run interactions, so the interaction coefficient is not fixed by construction. The paper also transparently acknowledges the measurement limitation that annual balance-sheet data cannot capture hidden insolvency, which is a data-quality concern rather than a circular argument. Self-citations to Correia, Luck and Verner (2025, 2026) are used for data construction and for external evidence from OCC examiners about failed-bank insolvency; these are checkable sources, not load-bearing circular assertions. Overall, no prediction in the paper reduces by definition to the fitted parameters, and the score reflects only the modest in-sample overlap between the fundamentals index and the failure outcome it is used to explain.
Assumptions & free parameters
free parameters (2)
- Bank fundamentals index coefficients =
Estimated from Table A.5 failure regressions
- Local projection lag length =
3 annual lags; 24 weekly lags
assumptions (4)
- domain assumption Newspaper articles provide a sufficiently complete record of bank runs.
- domain assumption Observable balance-sheet fundamentals proxy true bank health.
- domain assumption Local projection identification: run shocks are exogenous conditional on city fixed effects, lags, and controls.
- standard math Linear probability models are valid for binary outcomes.
Cite this review
Pith. "Pith review of Bank Runs With and Without Bank Failure." pith.science (2026). https://pith.science/paper/QS3FMZ32
@misc{pith2026260120285,
author = {Pith},
title = {Pith review of: Bank Runs With and Without Bank Failure},
year = {2026},
howpublished = {\url{https://pith.science/paper/QS3FMZ32}},
note = {Machine review of arXiv:2601.20285}
}
read the original abstract
We study the causes and consequences of bank runs. By applying large language models to historical newspapers, we create a comprehensive database of bank runs in U.S. history with information on 3,984 runs on individual banks from 1863 to 1934. Our novel data allow us to establish that runs are considerably more likely in weak banks but also occur in strong banks, especially in response to negative news about the real economy or the broader banking system. However, runs typically only result in failure for banks with poor fundamentals. Strong banks survive runs through various mechanisms, including signaling strength, interbank cooperation, and temporary suspension. At the local level, runs on banks with poor fundamentals translate into substantially larger declines in deposits, lending, and manufacturing activity than runs on strong banks. Our findings imply that poor fundamentals are central to explaining both when runs occur and when they have severe economic effects, tempering the view that small shocks can generate discontinuous jumps to bad equilibria through self-fulfilling run dynamics.
Figures
Forward citations
Cited by 1 Pith paper
-
Bank Run Exposure in a Paycheck-to-Paycheck Economy with Loss-Averse Depositors
A behavioral model shows that loss-averse, paycheck-to-paycheck depositors can trigger bank runs when they assign high probability to bad income states, and a Call Report exercise finds modest, imprecise empirical support.
Reference graph
Works this paper leans on
-
[1]
Geographic Names Information System (GNIS) Domestic Names dataset. After dropping locations other than cities and towns—ranging from ranches and towns to national parks and tribal areas—we end up with 209,257 cities and towns
-
[2]
IPUMS NHGIS place point GIS files (Schroeder et al., 2025). These “depict the locations of incorporated, unincorporated, and census-designated places for the entire U.S. from 1900 to 2015.” After processing this data and removing certain outliers—such as towns with impossibly long names—we obtain 22,844 cities and towns
work page 2025
-
[3]
Wikipedia articles on U.S. populated areas. We downloaded the entire list of Wikipedia articles, selected those about current and past U.S. populated areas, and compiled a list of those areas including their location and population across time. This yields 25,431 cities and towns
-
[4]
Census Bureau and Steiner, 2017)
Stanford’s Center for Spatial and Textual Analysis (U.S. Census Bureau and Steiner, 2017). This is a curated dataset of 8,848 cities and towns, mostly compiled from the U.S. census and state-level sources
work page 2017
-
[5]
Manually collected datasets. We further rely on the work of Jacob Alperin-Sheriff who manually digitized 20,844 cities from the U.S. decennial census, and of of James Feigenbaum, who digitized a partly overlapping set of 2,442 cities also from the census. When combining these cities, we perform minimal standardizations on city names, such as replacing “ce...
-
[6]
When combined, the previous sources are quite comprehensive but have some gaps
Lastly, a manually collected list of banks. When combined, the previous sources are quite comprehensive but have some gaps. For instance, they exclude state banks B.15 that operated exclusively between 1862 and 1869, as well as certain state banks, private banks, and trusts that operated between 1901 and 1940. To address this, we manually verify banks tha...
work page 1907
-
[36]
This contains a manually collected list of all U.S
Census of antebellum state banks (Weber, 2006). This contains a manually collected list of all U.S. state banks from 1782 to 1861
work page 2006
-
[37]
This contains the names and main events pertaining all national banks up to 1940
Census of national banks between 1863 and 1940 (Correia, Luck and Verner, 2025). This contains the names and main events pertaining all national banks up to 1940
work page 1940
Show all 41 references
-
[38]
These tables contain the names and main events about all commercial banks between approximately 1940 and today, plus some pre-1940 events for banks that still existed by that date
Federal Financial Institutions Examination Council (FFIEC) bank attribute tables. These tables contain the names and main events about all commercial banks between approximately 1940 and today, plus some pre-1940 events for banks that still existed by that date
1940
-
[39]
This dataset includes not only state banks but also private banks, not included in the previous sources
Census of all banks between 1870 and 1900 (Jaremski and Fishback, 2018). This dataset includes not only state banks but also private banks, not included in the previous sources
1900
-
[40]
Census of all banks regulated by the Department of Financial Services of the State of New York between 1784 and 2025 (New York State Department of Financial Services, 2024)
2025
-
[281]
First determine whether a run occurred at any point
-
[292]
Then determine whether a suspension occurred
-
[303]
Then determine whether the bank reopened or permanently closed
-
[314]
32 33These are the episode types: 34
Use this order to select the episode type. 32 33These are the episode types: 34
-
[341]
‘large‘: articles highlight a role for large depositors, businessmen, mercantile or commercial depositors, etc
-
[351]
‘run_only‘: A bank run on this bank took place, and the bank remains open throughout the entire episode (does not suspend or fail)
-
[352]
‘institutional‘: articles mention institutional depositors, such as county or city treasurers, trust funds, federal, universities, etc
-
[362]
As a consequence of the bank run, the bank suspends (either fully by suspending all payments or partially by invoking the 30/60/90 day rule)
‘run_suspension_reopening‘: A bank run takes place. As a consequence of the bank run, the bank suspends (either fully by suspending all payments or partially by invoking the 30/60/90 day rule). However, the bank eventually reopens and does not fail
-
[363]
‘minorities‘: the articles highlight the role of immigrant groups, specific ethnicities (japanese, german, hungarian, italian, etc.), freedmen, or other demographic groups distinguishable from the general public
-
[373]
As a consequence of the bank run, the bank suspends
‘run_suspension_closure‘: A bank run takes place. As a consequence of the bank run, the bank suspends. The bank remains closed permanently and has either a receiver assigned or is taken over by another bank
-
[374]
‘small‘: the articles highlight a role for small depositors or retail depositors
-
[384]
As a consequence of the bank run, the bank suspends
‘run_suspension_unsure‘: A bank run takes place. As a consequence of the bank run, the bank suspends. It is unsure if the bank remains closed permanently or if it reopens
-
[385]
‘workers‘: the articles highlight a role for specific professions such as farmers, miners, or factory workers
-
[395]
There is no bank run
‘suspension_closure‘: A bank suspends. There is no bank run. The bank remains closed permanently and has either a receiver assigned or is taken over by another bank
-
[396]
‘general‘: the article just mention the run was driven by the general B.12 public or general depositors
-
[406]
There is no bank run
‘suspension_reopening‘: A bank suspends. There is no bank run. The bank is able to reopen successfully and continues to operate its business
-
[407]
Colorado Territory,
‘unspecified‘: the articles only do not specify which type of depositors ran. 41 42 43# Guidelines 44 45- Only mark a category if the article describes the measure as a response to the bank run or depositor panic. 46- If the article does not clearly describe a measure being un...
-
[417]
As a consequence of the B.8 suspension, depositors run and/or are agitated (i.e., they gather outside the bank) in an ill-fated attempt to regain access to their funds
‘suspension_run_closure‘: A bank suspends. As a consequence of the B.8 suspension, depositors run and/or are agitated (i.e., they gather outside the bank) in an ill-fated attempt to regain access to their funds. The bank remains closed permanently and has either a receiver ass...
-
[428]
As a consequence of the suspension, depositors run and/or are agitated (i.e., they gather outside the bank) in an ill-fated attempt to regain access to their funds
‘suspension_run_reopening‘: A bank suspends. As a consequence of the suspension, depositors run and/or are agitated (i.e., they gather outside the bank) in an ill-fated attempt to regain access to their funds. The bank is able to reopen successfully and continues to operate it...
-
[439]
Explain in the ‘notes‘ field
‘other‘: Other type of event or sequence. Explain in the ‘notes‘ field. 44 45 46# Causes of bank runs and suspensions 47
-
[481]
‘rumor_or_misinformation‘: unfounded rumors or misinformation about the bank (false reports, misinterpreted events, malicious jokes, etc)
-
[492]
‘bank_specific_adverse_info‘: bank-specific adverse information affecting the solvency of the bank (embezzlement, suicides, discoveries of massive loans , or other scandals)
-
[503]
‘local_banks‘: information about runs and distress of other banks in the local economy
-
[514]
‘local_shock‘: information about the local economy that could affect the solvency of the bank (crop failures, default of local non-financial firms)
-
[525]
‘correspondent‘: failure or distress to correspondent banks or redemption agents
-
[536]
‘macro_news‘: systemic financial panics, nationwide downturns, or other macroeconomic or political shocks (wars, currency crises, etc)
-
[547]
Only applicable for suspensions, not for bank runs
‘government_action‘: closed by government action (by receiver, examiner, banking holiday, etc). Only applicable for suspensions, not for bank runs
-
[558]
Only applicable for suspensions, not for bank runs
‘voluntary_liquidation‘: voluntary liquidation and related events (charter expiration, etc). Only applicable for suspensions, not for bank runs
-
[562]
national
Output rules 57- Match found: If the article describes one or more relevant U.S. banking events, set ‘is_match‘ to True and return a list of event objects in the events field. Do not merge multiple events into a single object. Each event must be represented as its own object i...
1909
-
[569]
the Bank of Smyrna
‘other‘: other; unknown; or unable to tell. 57 58# Guidelines and notes 59 60You must follow these guidelines: 61 62- If you are unable to analyze the articles successfully, or if the article involves bank events in foreign countries outside of the U.S., set ‘success‘ to False...
Reviewed August 15, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.