Popular in Social Sciences

Political Science

V-Dem’s Conceptions of Democracy and Their Consequences

University of Notre Dame Michael Coppedge

Central Monetary Authorities

Macroeconomic modelling of CBDC: a critical review

European Central Bank Ulrich Bindseil

Social Identities and Divisions

A new explanation of antisemitism: Jew hatred as a civilisational phenomenon

Boston University Liah Greenfeld

Central Monetary Authorities

Financing Private Credit

Federal Reserve Bank of New York Nina Boyarchenko

Culture and Society

The City as a Growth Machine: Toward a Political Economy of Place

New York University Harvey Moltoch

American Politics

Keeping the Republic

Boston College Marc Landy

British Politics

Entrepreneurial thinking: the politics and practice of policy impact

University of Sheffield Matthew Flinders

Failing Banks

View more insights in Social Sciences - Economics - Financial Economics - Macroeconomics

View more figures from Massachusetts Institute of Technology

View the page of Emil Verner

Read the Study

Next in Economics

Fiscal policy volatility and capital misallocation

Shengyu Li | UNSW Sydney

Effects of monetary policy on household expectations: The role of homeownership

Shihan Xie | University of Illinois at Urbana-Champaign

Why do banks fail? Emil Verner’s comprehensive study of U.S. commercial banks from 1865 to 2023 points to three key factors: rising asset losses, deteriorating solvency, and increased reliance on expensive non-core funding. These elements are central to understanding bank failures throughout history.

Image courtesy of interviewee. September 26, 2024

Log-in or Sign-up to Faculti
Currently viewing this subject insight as a guest. You have insight(s) remaining for this month. Login to view 8000+ figures on the platform.

Generating Workspace...

Export as PDF

This video discusses the history and causes of bank failures, particularly focusing on the relationship between bank solvency and the occurrence of banking crises. It highlights two main explanations for bank failures: one is the "bank run" theory, which suggests that fear of losing deposits can trigger a run on a bank, leading to its demise even if it is initially solvent. The second explanation points to deteriorating solvency driven by factors such as credit risk and interest rate risk, which can cause banks to fail irrespective of any runs.

The documented evidence shows that bank failures due to widespread crises lead to significant macroeconomic downturns, as seen in historical events like the Great Depression and the 2008 financial crisis. The analysis emphasizes that failing banks typically exhibit warning signs years before their failure, including rising non-performing loans and increased reliance on expensive forms of funding.

The study further asserts that approximately 80% of bank failures can be attributed to insolvency rather than panic. Historical data supports this assertion, revealing that bank runs often respond to banks' poor fundamentals rather than being the root cause of failures. Thus, the authors argue that focusing on maintaining or restoring bank solvency should be a key strategy for policymakers to prevent and address bank failures, rather than solely preventing bank runs with liquidity support.

Language:

Export as PDF

Export as PDF
Here are 5 key points from the video:

1. **Connection Between Bank Failures and Economic Downturns**: The video emphasizes a strong historical correlation between banking crises characterized by widespread failures and severe macroeconomic downturns. Past episodes, such as the Great Depression and the 2008 financial crisis, are highlighted as examples where banking crises preceded significant economic contractions.

2. **Two Main Explanations for Bank Failures**: The document discusses two broad explanations for why banks fail: (a) bank runs, where depositors withdraw funds due to fears of insolvency; and (b) fundamental solvency issues, where deteriorating financial health leads to failure independent of runs. The importance of these factors in historical convideos is also analyzed.

3. **Predictability of Bank Failures**: The research indicates that many bank failures are highly predictable, often occurring following signals such as rising losses and deteriorating solvency, which can be detected years prior to the actual failure. This predictability suggests patterns that can be monitored over time.

4. **Role of Non-Core Funding**: Failing banks tend to increasingly rely on more expensive forms of funding—non-core funding—as they face difficulties. This shift increases their financing costs, which can further squeeze their margins and lead them to take on additional risks to maintain profitability.

5. **Policy Implications**: The findings underscore the necessity for policymakers to focus on bank solvency as a measure of stability, rather than merely preventing bank runs through liquidity backstops. This involves restoring and ensuring solvency, which could involve measures such as issuing more equity to strengthen banks' financial positions.

These points collectively highlight the complex interactions between bank solvency, economic health, and the preventive measures needed in banking policy.

Export as PDF
Here are some key quotes extracted from the video:

1. **On the urgency of addressing bank failures:**
- "There's, you know, by now a very large literature in macroeconomics and finance that documents and argues both theoretically and empirical that bank failures, and especially episodes of banking crises with widespread bank failures feature some of the most severe macroeconomic downturns."

2. **Historical convideo of banking crises:**
- "If you look at cross country data, going back over the past a hundred, 150 years, then times of banking crises where there's widespread bank failures are associated with some of the most severe recessions."

3. **Micro-level impact of bank distress:**
- "Firms and households that borrow from banks that are in distressed, they're the ones that contract investment the most during these downturns and employment the most."

4. **Two broad explanations for bank failures:**
- "Prominent accounts of major banking crises offer two broad explanations for why banks fail. One view is that bank failures are driven by bank runs."

5. **The nature of bank runs:**
- "Deposit outflows in the period right before failure of about 12% to 13%. These banks are losing a lot of deposits before failure."

6. **Deterioration of solvency before failure:**
- "The rising losses on their asset portfolios and deteriorating solvency many years before failure, typically failing banks start to get into trouble."

7. **On the patterns of failing banks:**
- "Failing banks often follow a boom and bust pattern...typically, banks would grow by trying to get new customers...that sort of predicts a poor performance by the bank down the road."

8. **The predictability of bank failures:**
- "These markers imply that bank failures are actually highly predictable throughout the historical sample."

9. **Solvency versus bank runs:**
- "In the history of the United States...most bank failures seem to be driven by deteriorating fundamentals that lead to insolvency."

10. **Policy implications of the findings:**
- "Policymakers shouldn't just prevent runs by providing sort of liquidity guarantees and backstops, but they should really be focused on ensuring that banks remain solvent."

These quotes capture the essence of the arguments presented in the video, which discusses the reasons behind bank failures, the predictability of these failures, and their implications for economic policy.

X-TIMESTAMP-MAP:LOCAL=00:00:00.000,MPEGTS=900000

00:00.000 --> 00:02.845
There's, you know, by now a very large literature in

00:02.845 --> 00:05.685
macroeconomics and finance that documents

00:05.685 --> 00:07.085
and argues both theoretically

00:07.085 --> 00:09.125
and empirical that bank failures,

00:09.305 --> 00:12.245
and especially episodes of banking crises

00:12.395 --> 00:15.165
with widespread bank failures feature some

00:15.165 --> 00:18.485
of the most severe macroeconomic, downturns.

00:18.705 --> 00:21.525
Um, so for example, if you look in sort

00:21.525 --> 00:23.205
of cross country data,

00:23.205 --> 00:26.605
going back over the past a hundred, 150 years, then times

00:26.745 --> 00:28.365
of, banking crises

00:28.415 --> 00:30.565
where there's widespread bank failures are associated

00:30.565 --> 00:32.845
with some of the most severe, recessions.

00:32.845 --> 00:35.485
So, example, you know, the Great Depression of the 1930s,

00:36.145 --> 00:39.645
the great recession of, of 2008 and other similar episodes.

00:40.145 --> 00:42.525
Um, and then at the micro level, we also know that,

00:42.785 --> 00:45.325
you know, firms and households that borrow from banks

00:45.325 --> 00:47.405
that are in distressed, they're the ones

00:47.405 --> 00:49.645
that contract investment the most,

00:49.645 --> 00:51.565
during these downturns and employment the most.

00:51.665 --> 00:53.125
So teams that banks sort

00:53.125 --> 00:55.565
of play a special role in providing capital.

00:55.665 --> 00:58.005
And when there's bank failures, then

00:58.005 --> 00:59.205
that role becomes impaired,

00:59.225 --> 01:01.365
and that contributes to making,

01:01.415 --> 01:03.365
these downturns very severe theory

01:03.425 --> 01:04.445
and prominent accounts

01:04.445 --> 01:07.365
of major banking crises offer two broad explanations

01:07.365 --> 01:08.405
for why banks fail.

01:08.945 --> 01:12.245
So one view is that bank failures are driven by bank runs.

01:12.705 --> 01:15.445
Um, so this, for example, shows up in the prominent work

01:15.445 --> 01:16.925
of Diamond and dfi,

01:16.925 --> 01:18.685
which won the Nobel Prize a few years ago.

01:19.025 --> 01:21.285
And it's this idea that depositors

01:21.305 --> 01:24.125
and bank creditors might run on the bank just

01:24.125 --> 01:26.125
because they fear that others will run on the bank.

01:26.465 --> 01:29.085
And when that bank run happens, the bank,

01:29.145 --> 01:31.805
is gonna do things that are, is gonna erode its solvency.

01:32.075 --> 01:35.005
Like, for example, fire sell assets and a,

01:35.005 --> 01:37.565
and a bank that was initially solvent then becomes

01:37.835 --> 01:39.165
insolvent because of that.

01:39.305 --> 01:43.125
Uh, run a second version of that bank run view is

01:43.125 --> 01:45.245
that if a bank is starting to get into trouble

01:45.265 --> 01:47.285
and has sort of deteriorating fundamentals,

01:47.755 --> 01:50.285
then depositors might start to worry about the bank

01:50.865 --> 01:54.455
and they'll run on the bank leading to the bank's demise,

01:55.155 --> 01:57.335
even though the bank actually was solvent

01:57.335 --> 01:59.335
and could have survived in the absence of the run.

01:59.635 --> 02:02.335
So that's sort of the first view, which is the bank run view

02:02.755 --> 02:03.775
of bank failures.

02:04.315 --> 02:05.775
The second view of bank failures,

02:05.775 --> 02:08.215
or the second reason why banks can fail, is

02:08.215 --> 02:11.455
because of deteriorating solvency That's gonna lead

02:11.455 --> 02:12.775
to failure irrespective

02:12.795 --> 02:15.015
of whether a bank run occurs, or not.

02:15.315 --> 02:16.695
So that can be because of, for example,

02:16.715 --> 02:18.135
the realization of credit risk.

02:18.475 --> 02:21.255
So loans that went bad that lead to losses for the banks,

02:21.635 --> 02:23.535
it could be a realization of interest rate, risk,

02:23.835 --> 02:24.855
or even things like, you know,

02:24.855 --> 02:26.575
governance issues and, and fraud.

02:26.835 --> 02:28.455
The literature has really,

02:28.515 --> 02:31.415
had a longstanding debate about the relative importance

02:31.415 --> 02:33.775
of these two channels for bank failures

02:33.775 --> 02:36.095
or reasons for bank failures, especially

02:36.095 --> 02:38.135
during major, banking crises.

02:38.755 --> 02:41.615
So, just to give some examples, if you look at, for example,

02:41.675 --> 02:43.895
the discussion on the 1930s, Friedman

02:43.895 --> 02:45.175
and Schwartz, for example, argued

02:45.175 --> 02:47.335
that panic runs were really important

02:47.395 --> 02:49.095
for understanding the widespread bank

02:49.095 --> 02:50.495
failures during the Great Depression.

02:51.045 --> 02:53.655
Then others like Wicker or Calomiris

02:53.655 --> 02:56.735
and Mason, Teman, other economic historians argue

02:56.765 --> 02:58.855
that actually many of the banks that failed

02:58.855 --> 03:00.365
during the Great Depression did so

03:00.365 --> 03:02.205
because they had really bad fundamentals

03:02.425 --> 03:04.245
and were essentially in insolvent.

03:04.585 --> 03:07.845
Uh, we saw similar debates in 2008 in understanding,

03:08.185 --> 03:10.445
you know, why the 2008 crisis was so bad

03:10.445 --> 03:13.525
and why there were so many, failures with some arguing

03:13.525 --> 03:14.725
that runs were very important

03:14.785 --> 03:18.125
and others arguing that runs were sort of a response

03:18.185 --> 03:19.605
to weak, fundamentals.

03:20.065 --> 03:22.485
And we saw it again with the failure of SVB.

03:22.865 --> 03:24.445
So there were some people that came out

03:24.445 --> 03:26.445
and said, look, this was a classic bank run

03:26.465 --> 03:30.125
and SVB, would've, failed, or failed

03:30.125 --> 03:31.405
because of the run,

03:31.405 --> 03:33.125
and would've survived in the absence of the run.

03:33.465 --> 03:34.605
And others argued, no.

03:34.605 --> 03:37.565
Look, SVB actually was a poorly run bank

03:37.565 --> 03:40.445
that had fundamental, problems exposure

03:40.465 --> 03:42.445
to interest rate risk that made it insolvent

03:42.785 --> 03:45.605
and effectively the bank, you know, would've failed anyway.

03:45.625 --> 03:48.485
It wasn't the run. While the run may have determined exactly

03:48.485 --> 03:50.565
when the bank failed, it wasn't the run itself,

03:50.985 --> 03:52.205
that led to the failure.

03:52.825 --> 03:56.125
And so, the goal of our paper was to go back

03:56.125 --> 03:57.965
and study the nearly complete history

03:58.305 --> 04:01.685
of bank failures in the United States since the Civil War,

04:01.745 --> 04:06.085
so since 18 60, 65, to understand why banks fail.

04:06.625 --> 04:10.085
Um, and so we do this using this long historical sample

04:10.475 --> 04:13.045
because this is kind of a nice laboratory

04:13.285 --> 04:15.365
where we have a period of US history

04:15.905 --> 04:18.165
before the Federal Reserve system,

04:18.165 --> 04:20.565
which is introduced in 1914, and

04:20.565 --> 04:23.725
before deposit insurance, which comes in in 1934.

04:24.065 --> 04:26.605
So we have a period when essentially governments aren't,

04:27.065 --> 04:29.605
you know, can't play the role of preventing failures

04:29.605 --> 04:30.845
through deposit insurance

04:31.185 --> 04:33.805
or a lender of last resort activity.

04:33.905 --> 04:36.205
So there's no sort of government, safety net.

04:36.425 --> 04:38.085
And then we can look at the more modern period

04:38.085 --> 04:40.125
where we have all of these safety nets

04:40.125 --> 04:41.805
and try to see what's the role

04:41.985 --> 04:45.485
of bank runs versus fundamental solvency problems in

04:45.485 --> 04:48.725
explaining failures in sort of this broad sweep of,

04:48.745 --> 04:51.925
of US history to do this comprehensive analysis

04:52.025 --> 04:54.925
of bank failures in the US going back to the Civil War.

04:55.385 --> 04:57.925
We can't just use existing, data sets.

04:57.985 --> 05:00.405
Um, there isn't a data set you can download that says,

05:00.705 --> 05:02.725
you know, what, what are the conditions of banks,

05:03.185 --> 05:04.325
going that far back?

05:04.425 --> 05:07.845
So we had to construct a new database,

05:07.845 --> 05:10.005
that covers most commercial banks in the

05:10.025 --> 05:11.645
us over this period.

05:12.145 --> 05:13.485
So for the historical period

05:13.505 --> 05:15.205
before World War ii,

05:15.205 --> 05:18.365
we used newly digitized data on national banks,

05:18.455 --> 05:20.525
which is a subset of the US banking system.

05:20.545 --> 05:22.445
So there's national banks and state banks,

05:22.505 --> 05:24.205
and we cover, national banks,

05:24.205 --> 05:27.325
which typically have a market share of say, you know, 40

05:27.345 --> 05:29.525
to 60% of the banking system.

05:29.985 --> 05:33.245
And so we construct basically fundamental information on

05:33.245 --> 05:35.165
these national banks every year.

05:35.165 --> 05:37.085
So we have their balance sheets,

05:37.105 --> 05:39.445
and we see whether the banks, fail

05:39.745 --> 05:42.805
and we see information on the types of,

05:42.905 --> 05:45.005
or the state of their banks in failure.

05:46.275 --> 05:48.215
And, we do this,

05:48.215 --> 05:51.055
using some newly developed optical character recognition

05:51.055 --> 05:54.335
methods to digitize existing PDFs,

05:54.335 --> 05:56.855
and turn these into, you know, databases

05:56.855 --> 05:58.325
That you could analyze in Excel

05:58.665 --> 06:00.365
or using statistical software.

06:00.825 --> 06:02.965
Um, then for the modern period,

06:03.065 --> 06:06.725
we use data on commercial banks from the call reports,

06:06.775 --> 06:10.365
which is the standard source, for data on balance sheets

06:10.365 --> 06:12.685
and income statements from, commercial bank.

06:13.225 --> 06:16.125
So this gives us essentially data on the fundamentals

06:16.145 --> 06:20.245
of most banks in the US for this, you know, roughly 100

06:20.305 --> 06:21.685
and and 60 year period.

06:22.465 --> 06:24.565
Um, on top of that, we have a little bit more data.

06:24.705 --> 06:29.125
So we also have data on the perceived cause of failure

06:29.825 --> 06:33.365
by bank examiners, for the historical sample.

06:33.505 --> 06:35.085
So before World War ii.

06:35.545 --> 06:37.005
So, the bank examiners

06:37.005 --> 06:39.565
that would go into failing banks from the OCC,

06:39.565 --> 06:42.565
which was the agency that, that, regulated these banks,

06:42.915 --> 06:44.845
they would go in and give a cause of failure.

06:44.945 --> 06:46.325
And so we could see well, on the ground,

06:46.475 --> 06:50.085
what did the bank examiners think was the cause of failure.

06:50.625 --> 06:52.925
And then we can also see information like what's the

06:53.125 --> 06:55.565
recovery rate on the assets of the banks that failed,

06:55.585 --> 06:58.605
and how much did depositors lose in failing banks?

06:58.945 --> 07:01.525
So this gives us a quite comprehensive view of,

07:01.625 --> 07:02.645
of bank failures.

07:03.225 --> 07:05.125
Um, and with this dataset,

07:05.145 --> 07:08.565
we outline three basic facts about the commonalities

07:08.625 --> 07:09.685
in failing banks.

07:09.985 --> 07:12.925
So first, we show that failing banks,

07:13.425 --> 07:16.845
the rising losses on their asset portfolios

07:17.225 --> 07:20.085
and deteriorating solvency many years

07:20.085 --> 07:22.485
before failure, typically failing banks

07:22.485 --> 07:23.525
start to get into trouble.

07:23.625 --> 07:24.765
You know, five, six years

07:24.765 --> 07:27.605
before failure, their loan book, has, you know,

07:27.605 --> 07:29.005
rising non-performing loans

07:29.585 --> 07:33.045
and their equity position gets, compressed,

07:33.045 --> 07:34.765
and their capitalization deteriorates

07:34.765 --> 07:36.445
because of these losses.

07:36.985 --> 07:39.045
So in other words, these failures don't just sort

07:39.045 --> 07:42.685
of come out of the blue in terms of, in, you know, in,

07:42.705 --> 07:44.285
in terms of the fundamentals.

07:44.485 --> 07:46.125
Actually the fundamentals are sent sending a signal

07:46.125 --> 07:48.205
that these banks have quite,

07:48.525 --> 07:49.965
deteriorating, solvency.

07:51.035 --> 07:52.935
The second, fact is

07:52.935 --> 07:57.015
that failing banks across this entire sample typically end

07:57.015 --> 07:59.975
up having to rely more on expensive forms of

08:00.085 --> 08:01.775
what we call non-core funding.

08:02.315 --> 08:04.135
So non-core funding is essentially

08:04.135 --> 08:05.615
non-retail deposit funding.

08:05.835 --> 08:08.735
So think market-based funding, wholesale funding

08:09.235 --> 08:10.575
or broker deposits.

08:10.705 --> 08:12.015
These are types of funding

08:12.015 --> 08:13.895
that are much more expensive relative

08:13.895 --> 08:15.935
to very low interest deposit funding.

08:16.435 --> 08:18.615
And what that means is that once banks have

08:18.615 --> 08:20.855
to start relying on this type of funding, then

08:20.855 --> 08:22.735
that's gonna increase their interest cost

08:23.115 --> 08:24.935
and potentially squeeze their margins

08:24.935 --> 08:27.655
or force them to have to take more risk in order

08:27.655 --> 08:29.575
to maintain their interest margins.

08:29.575 --> 08:32.255
And so on the, on the funding cost side,

08:32.355 --> 08:35.815
you see a deteriorating performance, in failing banks.

08:36.585 --> 08:38.525
And then the third basic fact is

08:38.525 --> 08:42.365
that failing banks often follow a boom and bust pattern.

08:42.865 --> 08:46.805
So one of the main reasons why banks get into trouble down

08:46.805 --> 08:49.725
the road is because they grew very quickly in the past.

08:50.185 --> 08:53.285
Um, so typically, you know, banks would grow by trying

08:53.285 --> 08:56.365
to get new customers, making lots of loans,

08:56.365 --> 08:58.085
especially real estate based loans.

08:58.745 --> 09:02.085
And that, you know, boom leads the bank, sometimes

09:02.105 --> 09:03.325
to overextend itself.

09:03.585 --> 09:05.525
And that sort of predicts,

09:05.925 --> 09:07.325
a poor performance by the bank.

09:07.625 --> 09:11.325
Um, down the road we see in failing banks these,

09:11.765 --> 09:14.685
patterns of deteriorating solvency,

09:14.705 --> 09:17.565
rising reliance on non-core and expensive forms of funding

09:17.705 --> 09:20.285
and this boom and bust pattern in the decade before failure.

09:20.945 --> 09:22.845
And so it turns out what this means is

09:22.845 --> 09:26.925
that these markers are actually a strong signal,

09:27.025 --> 09:29.245
of impending failure, for banks.

09:29.745 --> 09:33.565
Um, so actually it turns out that these markers imply that

09:34.275 --> 09:37.085
bank failures are actually highly predictable,

09:37.095 --> 09:39.125
throughout the historical, sample.

09:39.545 --> 09:42.725
So a statistician using this information would be able

09:42.725 --> 09:45.325
to sort of detect, failure rates with a,

09:45.365 --> 09:47.565
a very favorable sort of true positive relative

09:47.565 --> 09:50.405
to false positive sort of, sort of trade off.

09:50.625 --> 09:53.125
And this holds, even in the historical sample

09:53.585 --> 09:56.525
before deposit insurance, when one might have thought, well,

09:56.525 --> 09:59.245
maybe there were some failures that were driven by sort

09:59.245 --> 10:02.325
of unexpected runs or unexpected, problems.

10:02.825 --> 10:06.005
Um, and, and this predictability is quite strong

10:06.465 --> 10:08.965
and it matters not just for individual bank failures,

10:09.625 --> 10:12.565
it also matters for aggregate, bank failures.

10:12.585 --> 10:14.165
So waves of bank failures

10:14.345 --> 10:17.845
during the major banking crises in the us like the panic

10:17.905 --> 10:21.245
of 1873 or the panic of 1893

10:21.785 --> 10:24.285
or the panic of 1907, the Great Depression,

10:24.585 --> 10:26.565
the SNL crisis, the 2008 crisis.

10:27.055 --> 10:28.285
These are all

10:28.465 --> 10:32.445
to a substantial extent predictable based on deteriorating

10:32.445 --> 10:35.245
bank fundamentals in the period, before.

10:35.425 --> 10:37.285
So bank failures are very predictable,

10:37.285 --> 10:39.805
both in the historical and the, and the modern sample.

10:40.345 --> 10:41.485
In the historical sample.

10:41.545 --> 10:45.085
We can also specifically analyze failures with bank runs.

10:45.585 --> 10:47.965
So these are the types of episodes that, you know, are,

10:47.985 --> 10:50.525
are famous from movies like It's A Wonderful Life

10:50.545 --> 10:53.725
or Mary Poppins episodes where depositors show up

10:54.145 --> 10:56.325
and, you know, wanna withdraw from from the bank.

10:56.575 --> 10:58.805
These types of failures are very common

10:59.145 --> 11:02.485
before deposit insurance is introduced, in 1934.

11:02.945 --> 11:05.165
In fact, we find that the typical bank failure

11:05.345 --> 11:09.605
before deposit insurance saw deposit outflows in the period

11:09.615 --> 11:12.765
right before failure of about 12 13%.

11:12.855 --> 11:15.565
These banks are losing a lot of deposits before failure.

11:16.365 --> 11:18.005
Nevertheless, we show

11:18.005 --> 11:20.445
that despite the banks experiencing runs,

11:21.135 --> 11:23.405
these failures are actually very predictable

11:23.425 --> 11:24.445
as well in the sense

11:24.445 --> 11:27.925
that these banks also have very poor fundamentals in the

11:27.975 --> 11:29.885
runup, to, to failure.

11:30.465 --> 11:33.525
Uh, and what this means is that these failures with runs,

11:33.875 --> 11:36.085
they're not sort of happening out of the blue

11:36.275 --> 11:38.365
because of some unfounded concerns

11:38.365 --> 11:39.405
about, you know, these banks.

11:39.905 --> 11:42.445
It suggests that more often than not,

11:42.665 --> 11:44.925
the runs on these banks are really a response

11:44.945 --> 11:46.525
to the weakness in the banks,

11:46.625 --> 11:48.325
the poor fundamentals in the bank,

11:48.705 --> 11:51.205
rather than the root cause of failure.

11:51.505 --> 11:52.725
We argue that sort of our,

11:52.785 --> 11:56.645
Our statistical interpretation that these failures

11:56.645 --> 11:59.805
with runs aren't driven per se by the runs themselves.

12:00.345 --> 12:01.845
We argue that that's actually consistent

12:01.845 --> 12:04.165
with historical accounts of these failures.

12:04.585 --> 12:06.005
So for example, as I mentioned

12:06.005 --> 12:09.965
before, we have data on OCC bank examiners classifications

12:10.625 --> 12:12.165
of why they thought the bank failed.

12:12.305 --> 12:14.245
And in the vast majority of failures,

12:14.465 --> 12:17.765
the bank examiners assign the cause of failure to be losses.

12:18.625 --> 12:21.005
Uh, an economic downturn,

12:21.025 --> 12:24.605
or fraud, by the bank directors runs account

12:24.625 --> 12:27.565
for less than 2% of the, of their assigned failures.

12:27.595 --> 12:30.125
Even though many of the failures featured runs

12:30.225 --> 12:33.845
or coincided, with runs and above

12:33.845 --> 12:36.805
and beyond that, we see that, you know, even in the failures

12:36.805 --> 12:40.845
with runs the recovery rates on the assets in those banks.

12:41.025 --> 12:43.205
So if you look at the assets that are left in those banks,

12:43.615 --> 12:45.925
those assets tend to have very poor quality.

12:46.385 --> 12:47.445
So often these are loans

12:47.445 --> 12:50.445
that actually the banks are not gonna collect very much on.

12:50.865 --> 12:52.845
Uh, and that suggests

12:52.845 --> 12:55.845
that really the depositors are responding to insolvency

12:55.845 --> 12:59.085
to weak fundamentals rather than depositors leading

12:59.145 --> 13:00.725
to failure of banks

13:00.725 --> 13:02.965
that would've survived in the absence of the run.

13:03.165 --> 13:06.925
A broad conclusion in terms of why banks fail is

13:06.925 --> 13:08.885
that in the history of the United States,

13:09.435 --> 13:12.605
when there have been lots of bank failures, so in fact,

13:12.665 --> 13:17.205
our data contains over 4,700 bank failures,

13:17.745 --> 13:21.005
we see that actually most bank failures seem to be driven

13:21.025 --> 13:24.245
by deteriorating fundamentals that lead to insolvency.

13:24.785 --> 13:28.645
So we argue that roughly 80% of bank failures even

13:28.645 --> 13:31.285
before deposit insurance, can be explained

13:31.285 --> 13:32.485
by insolvency alone.

13:32.965 --> 13:34.885
Moreover, you know, the banking crises

13:34.995 --> 13:36.885
that feature widespread bank failures,

13:37.235 --> 13:39.845
they should be thought of as solvency crises typically

13:39.995 --> 13:42.685
that follow a period of risk taking and rapid loan growth.

13:43.075 --> 13:46.285
They shouldn't be thought of, as coming from panics

13:46.435 --> 13:49.605
that arise sort of for no good fundamental, reasons.

13:50.425 --> 13:53.365
So related to that, we argue that sort of these types

13:53.365 --> 13:56.325
of failures, this story that non fundamental runs,

13:56.465 --> 13:59.565
so bank runs based on some rumor that's unfounded.

13:59.945 --> 14:01.925
We argue that those are might happen,

14:02.105 --> 14:04.925
but they're not a common cause of bank, failure.

14:05.805 --> 14:09.125
Moreover, even sort of bank runs that are driven

14:09.185 --> 14:12.285
by weak fundamentals, they can explain some failures,

14:12.825 --> 14:14.165
but actually we argue that in,

14:14.185 --> 14:16.645
in the data most runs are really a response

14:16.645 --> 14:18.325
to the four four fundamentals

14:18.505 --> 14:20.925
and the bank would likely have failed at some point.

14:21.385 --> 14:22.445
Uh, anyway,

14:23.045 --> 14:26.525
moreover, it looks like runs actually seem

14:26.525 --> 14:29.605
to happen a bit later than we would've expected based on

14:29.605 --> 14:31.845
theoretical benchmarks of kind of rational

14:31.845 --> 14:33.085
and attentive depositors.

14:33.625 --> 14:36.245
And the reason that our paper argues this is

14:36.245 --> 14:38.685
that we observe in the data banks that fail,

14:39.115 --> 14:41.325
that have very high probabilities of failure

14:41.785 --> 14:43.245
for several years before failure.

14:43.245 --> 14:45.245
Sometimes it's exceeding 20%.

14:45.905 --> 14:48.605
So imagine if you are fully rational

14:48.665 --> 14:51.085
and you know that your bank has a one in five chance

14:51.165 --> 14:53.885
of failing, and this is before deposit insurance.

14:53.945 --> 14:57.285
So if the bank fails, you lose 30 to 35%

14:57.545 --> 14:58.885
of your money, right?

14:59.115 --> 15:02.045
Then you should run immediately if you are completely

15:02.045 --> 15:03.085
attentive and aware.

15:03.385 --> 15:04.565
But it seems like in the data

15:04.835 --> 15:06.205
that actually often doesn't happen,

15:06.205 --> 15:09.285
or it happens more slowly than we might expect consistent

15:09.285 --> 15:10.565
with this idea that, you know,

15:10.565 --> 15:14.365
depositors are often not paying attention, somewhat sleepy.

15:14.705 --> 15:17.125
And, you know, they only wake up relatively late

15:17.175 --> 15:19.285
after problems in banks,

15:19.285 --> 15:21.045
become sort of very, very apparent.

15:21.585 --> 15:25.125
Uh, and so more broadly, I think kind of the conclusions

15:25.125 --> 15:28.005
and implications of our study are that, you know,

15:28.485 --> 15:31.685
focusing on solvency from a policy perspective,

15:32.155 --> 15:35.085
both ex andante to try to prevent bank bank failures

15:35.505 --> 15:39.045
and ex post during crises to, you know, mitigate crises.

15:39.845 --> 15:43.005
Policymakers shouldn't just prevent runs by providing sort

15:43.005 --> 15:44.845
of liquidity guarantees and backstops,

15:44.845 --> 15:47.325
but they should really be focused on ensuring

15:47.355 --> 15:49.485
that banks remain solvent

15:49.865 --> 15:53.565
or, restore solvency by issuing, more equity.

"Understanding Bank Failures: Historical Insights and Theoretical Perspectives"**

00:00:00 - There's, you know, by now a very large literature in macroeconomics and finance that documents and argues both theoretically and empirical that bank failures, and especially episodes of banking crises with widespread bank failures feature some of the most severe macroeconomic, downturns. so for example, if you look in sort of cross country data, going back over the past a hundred, 150 years, then times of, banking crises where there's widespread bank failures are associated with some of the most severe, recessions. So, example, you know, the Great Depression of the 1930s, the great recession of, of 2008 and other similar episodes. and then at the micro level, we also know that, you know, firms and households that borrow from banks
00:00:47 - that contract investment the most, during these downturns and employment the most. So teams that banks sort of play a special role in providing capital. And when there's bank failures, then that role becomes impaired, and that contributes to making, these downturns very severe theory and prominent accounts of major banking crises offer two broad explanations for why banks fail. So one view is that bank failures are driven by bank runs. so this, for example, shows up in the prominent work of Diamond and dfi, which won the Nobel Prize a few years ago. And it's this idea that depositors and bank creditors might run on the bank just because they fear that others will run on the bank.
00:01:29 - is gonna do things that are, is gonna erode its solvency. Like, for example, fire sell assets and a, and a bank that was initially solvent then becomes insolvent because of that. run a second version of that bank run view is that if a bank is starting to get into trouble and has sort of deteriorating fundamentals, then depositors might start to worry about the bank and they'll run on the bank leading to the bank's demise, even though the bank actually was solvent and could have survived in the absence of the run. So that's sort of the first view, which is the bank run view of bank failures. The second view of bank failures, or the second reason why banks can fail, is
00:02:11 - to failure irrespective of whether a bank run occurs, or not. So that can be because of, for example, the realization of credit risk. So loans that went bad that lead to losses for the banks, it could be a realization of interest rate, risk, or even things like, you know, governance issues and, and fraud. The literature has really, had a longstanding debate about the relative importance of these two channels for bank failures or reasons for bank failures, especially during major, banking crises. So, just to give some examples, if you look at, for example, the discussion on the 1930s, Friedman and Schwartz, for example, argued that panic runs were really important for understanding the widespread bank
00:02:51 - Then others like Wicker or Calomiris and Mason, Teman, other economic historians argue that actually many of the banks that failed during the Great Depression did so because they had really bad fundamentals and were essentially in insolvent. we saw similar debates in 2008 in understanding, you know, why the 2008 crisis was so bad and why there were so many, failures with some arguing that runs were very important and others arguing that runs were sort of a response to weak, fundamentals. And we saw it again with the failure of SVB. So there were some people that came out and said, look, this was a classic bank run and SVB, would've, failed, or failed
00:03:31 - and would've survived in the absence of the run. And others argued, no. Look, SVB actually was a poorly run bank that had fundamental, problems exposure to interest rate risk that made it insolvent and effectively the bank, you know, would've failed anyway. It wasn't the run. While the run may have determined exactly when the bank failed, it wasn't the run itself, that led to the failure. And so, the goal of our paper was to go back and study the nearly complete history of bank failures in the United States since the Civil War, so since 18 60, 65, to understand why banks fail. and so we do this using this long historical sample

"Analyzing Bank Failures: Historical Perspectives and Methodological Innovations"

00:04:13 - where we have a period of US history before the Federal Reserve system, which is introduced in 1914, and before deposit insurance, which comes in in 1934. So we have a period when essentially governments aren't, you know, can't play the role of preventing failures through deposit insurance or a lender of last resort activity. So there's no sort of government, safety net. And then we can look at the more modern period where we have all of these safety nets and try to see what's the role of bank runs versus fundamental solvency problems in explaining failures in sort of this broad sweep of, of US history to do this comprehensive analysis
00:04:55 - We can't just use existing, data sets. there isn't a data set you can download that says, you know, what, what are the conditions of banks, going that far back? So we had to construct a new database, that covers most commercial banks in the us over this period. So for the historical period before World War ii, we used newly digitized data on national banks, which is a subset of the US banking system. So there's national banks and state banks, and we cover, national banks, which typically have a market share of say, you know, 40 to 60% of the banking system. And so we construct basically fundamental information on
00:05:35 - So we have their balance sheets, and we see whether the banks, fail and we see information on the types of, or the state of their banks in failure. And, we do this, using some newly developed optical character recognition methods to digitize existing PDFs, and turn these into, you know, databases That you could analyze in Excel or using statistical software. then for the modern period, we use data on commercial banks from the call reports, which is the standard source, for data on balance sheets and income statements from, commercial bank. So this gives us essentially data on the fundamentals
00:06:20 - and and 60 year period. on top of that, we have a little bit more data. So we also have data on the perceived cause of failure by bank examiners, for the historical sample. So before World War ii. So, the bank examiners that would go into failing banks from the OCC, which was the agency that, that, regulated these banks, they would go in and give a cause of failure. And so we could see well, on the ground, what did the bank examiners think was the cause of failure. And then we can also see information like what's the recovery rate on the assets of the banks that failed, and how much did depositors lose in failing banks?
00:07:01 - of bank failures. and with this dataset, we outline three basic facts about the commonalities in failing banks. So first, we show that failing banks, the rising losses on their asset portfolios and deteriorating solvency many years before failure, typically failing banks start to get into trouble. You know, five, six years before failure, their loan book, has, you know, rising non-performing loans and their equity position gets, compressed, and their capitalization deteriorates because of these losses. So in other words, these failures don't just sort
00:07:42 - in terms of the fundamentals. Actually the fundamentals are sent sending a signal that these banks have quite, deteriorating, solvency. The second, fact is that failing banks across this entire sample typically end up having to rely more on expensive forms of what we call non-core funding. So non-core funding is essentially non-retail deposit funding. So think market-based funding, wholesale funding or broker deposits. These are types of funding that are much more expensive relative to very low interest deposit funding. And what that means is that once banks have to start relying on this type of funding, then

"Predicting Bank Failures: Analyzing Patterns of Solvency Deterioration and the Impact of Funding Practices"

00:08:23 - and potentially squeeze their margins or force them to have to take more risk in order to maintain their interest margins. And so on the, on the funding cost side, you see a deteriorating performance, in failing banks. And then the third basic fact is that failing banks often follow a boom and bust pattern. So one of the main reasons why banks get into trouble down the road is because they grew very quickly in the past. so typically, you know, banks would grow by trying to get new customers, making lots of loans, especially real estate based loans. And that, you know, boom leads the bank, sometimes
00:09:03 - And that sort of predicts, a poor performance by the bank. down the road we see in failing banks these, patterns of deteriorating solvency, rising reliance on non-core and expensive forms of funding and this boom and bust pattern in the decade before failure. And so it turns out what this means is that these markers are actually a strong signal, of impending failure, for banks. so actually it turns out that these markers imply that bank failures are actually highly predictable, throughout the historical, sample. So a statistician using this information would be able
00:09:45 - a very favorable sort of true positive relative to false positive sort of, sort of trade off. And this holds, even in the historical sample before deposit insurance, when one might have thought, well, maybe there were some failures that were driven by sort of unexpected runs or unexpected, problems. and, and this predictability is quite strong and it matters not just for individual bank failures, it also matters for aggregate, bank failures. So waves of bank failures during the major banking crises in the us like the panic of 1873 or the panic of 1893 or the panic of 1907, the Great Depression,
00:10:27 - These are all to a substantial extent predictable based on deteriorating bank fundamentals in the period, before. So bank failures are very predictable, both in the historical and the, and the modern sample. In the historical sample. We can also specifically analyze failures with bank runs. So these are the types of episodes that, you know, are, are famous from movies like It's A Wonderful Life or Mary Poppins episodes where depositors show up and, you know, wanna withdraw from from the bank. These types of failures are very common before deposit insurance is introduced, in 1934. In fact, we find that the typical bank failure
00:11:09 - right before failure of about 12 13%. These banks are losing a lot of deposits before failure. Nevertheless, we show that despite the banks experiencing runs, these failures are actually very predictable as well in the sense that these banks also have very poor fundamentals in the runup, to, to failure. and what this means is that these failures with runs, they're not sort of happening out of the blue because of some unfounded concerns about, you know, these banks. It suggests that more often than not, the runs on these banks are really a response to the weakness in the banks,
00:11:48 - rather than the root cause of failure. We argue that sort of our, Our statistical interpretation that these failures with runs aren't driven per se by the runs themselves. We argue that that's actually consistent with historical accounts of these failures. So for example, as I mentioned before, we have data on OCC bank examiners classifications of why they thought the bank failed. And in the vast majority of failures, the bank examiners assign the cause of failure to be losses. an economic downturn, or fraud, by the bank directors runs account for less than 2% of the, of their assigned failures.

"Understanding Bank Failures: The Crucial Role of Solvency Over Panic Runs"

00:12:30 - or coincided, with runs and above and beyond that, we see that, you know, even in the failures with runs the recovery rates on the assets in those banks. So if you look at the assets that are left in those banks, those assets tend to have very poor quality. So often these are loans that actually the banks are not gonna collect very much on. and that suggests that really the depositors are responding to insolvency to weak fundamentals rather than depositors leading to failure of banks that would've survived in the absence of the run. A broad conclusion in terms of why banks fail is that in the history of the United States,
00:13:12 - our data contains over 4,700 bank failures, we see that actually most bank failures seem to be driven by deteriorating fundamentals that lead to insolvency. So we argue that roughly 80% of bank failures even before deposit insurance, can be explained by insolvency alone. Moreover, you know, the banking crises that feature widespread bank failures, they should be thought of as solvency crises typically that follow a period of risk taking and rapid loan growth. They shouldn't be thought of, as coming from panics that arise sort of for no good fundamental, reasons.
00:13:53 - of failures, this story that non fundamental runs, so bank runs based on some rumor that's unfounded. We argue that those are might happen, but they're not a common cause of bank, failure. Moreover, even sort of bank runs that are driven by weak fundamentals, they can explain some failures, but actually we argue that in, in the data most runs are really a response to the four four fundamentals and the bank would likely have failed at some point. anyway, moreover, it looks like runs actually seem to happen a bit later than we would've expected based on theoretical benchmarks of kind of rational
00:14:33 - And the reason that our paper argues this is that we observe in the data banks that fail, that have very high probabilities of failure for several years before failure. Sometimes it's exceeding 20%. So imagine if you are fully rational and you know that your bank has a one in five chance of failing, and this is before deposit insurance. So if the bank fails, you lose 30 to 35% of your money, right? Then you should run immediately if you are completely attentive and aware. But it seems like in the data that actually often doesn't happen, or it happens more slowly than we might expect consistent with this idea that, you know, depositors are often not paying attention, somewhat sleepy.
00:15:49 - after problems in banks, become sort of very, very apparent. and so more broadly, I think kind of the conclusions and implications of our study are that, you know, focusing on solvency from a policy perspective, both ex andante to try to prevent bank bank failures and ex post during crises to, you know, mitigate crises. Policymakers shouldn't just prevent runs by providing sort of liquidity guarantees and backstops, but they should really be focused on ensuring that banks remain solvent or, restore solvency by issuing, more equity.

Faculti AI can make mistakes

Research

Policy

Analysis