View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

w o r k i n g
p

a

p

e

r

14 12

Tracing Out Capital Flows:
How Financially Integrated Banks
Respond to Natural Disasters
Kristle Cortés and Philip E. Strahan

FEDERAL RESERVE BANK OF CLEVELAND

Working papers of the Federal Reserve Bank of Cleveland are preliminary materials circulated to
stimulate discussion and critical comment on research in progress. They may not have been subject to the
formal editorial review accorded official Federal Reserve Bank of Cleveland publications. The views stated
herein are those of the authors and are not necessarily those of the Federal Reserve Bank of Cleveland or of
the Board of Governors of the Federal Reserve System.
Working papers are available at:

www.clevelandfed.org/research.

Working Paper 14-12

September 2014

Tracing Out Capital Flows:
How Financially Integrated Banks Respond to Natural Disasters
Kristle Cortés and Philip E. Strahan

Multi-market banks reallocate capital when local credit demand increases after natural disasters.
Following such events, credit in unaffected but connected markets declines by about 50 cents per
dollar of additional lending in shocked areas, but most of the decline comes from loans in areas
where banks do not own branches. Moreover, banks increase sales of more-liquid loans in order
to lessen the impact of the demand shock on credit supply. Larger, multi-market banks appear
better able than smaller ones to shield credit supplied to their core markets (those with branches)
by aggressively cutting back lending outside those markets.

JEL Codes: G20, G21.
Keywords: Financial Integration, Branch Banking, Securitization.

Suggested citation: Cortés, Kristle, and Philip E. Strahan, “Tracing Out Capital Flows: How
Financially Integrated Banks Respond to Natural Disasters,” Federal Reserve Bank of Cleveland,
working paper no. 14-12.

Kristle Cortés is at the Federal Reserve Bank of Cleveland and can be reached at kristle.
romerocortes@clev.frb.org. Philip E. Strahan is at Boston College and NBER.

1. INTRODUCTION
This paper traces out how multi-market banks alter their credit supply decisions in
response to local, exogenous shocks to credit demand stimulated by natural disasters. We find
that financially integrated banks reallocate funds toward markets with high credit demand and
away from other markets (“connected markets”) in which they lend. Thus, credit seems to flow
within banks toward high-return markets (where demand is high), and away from lower-return
ones within banking organizations. On average, credit supplied by banks to connected markets
declines by about 50 cents per dollar of increased lending in their shocked areas.
Why might shocks to loan demand affect credit supplied to connected markets by banks
exposed to disasters? The short answer is that lenders must finance higher loan demand. The
longer answer: banks face frictions in accessing external financing, both because issuing
additional deposits and raising equity is expensive, and because selling loans to third parties is
often difficult or impossible. Hence, exposed banks may have to cut lending in connected
markets to have the balance sheet capacity to accommodate higher demand. Whether or not the
decline lowers overall credit supply in connected markets depends on the presence of a second,
asset-side friction. If some of the displaced lending is to borrowers over which the affected bank
has a cost advantage relative to competing banks, then aggregate supply would fall. Absent such
advantages, other banks could step in to replace the connected bank, or the connected bank could
sell loans to third parties.
We find that the decline in lending in connected markets is concentrated outside of
banks’ core markets, which we define as those where the bank has a branch presence. Existing
evidence suggests that a bank’s physical presence in a market improves access to information

1

about borrower quality and the value of collateral (Berger et al., 2005; Degryse and Ongena,
2005; Loutskina and Strahan, 2009; Agrawal and Hauswald, 2010; Ergunor, 2010; Cortés, 2012).
Better than average access to local information can allow banks to earn rents, but also erects a
barrier to loan sales and/or securitization.1 Our findings suggest that banks protect rents that they
are able to earn in their core markets (or more correctly quasi-rents, as it is costly to build a local
infrastructure), by cutting lending sharply in markets where their ability to generate rents are less
important (i.e. markets where they lend without a physical presence). Since other lenders can
replace the lost credit in these non-core markets, where the exposed banks have no particular
informational or cost advantage, aggregate effects on credit supply to connected markets are
likely to be small.
We exploit natural disasters – hurricanes, earthquakes, tornadoes, floods, etc. – to
generate exogenous increases in local credit demand, and test how these increases in demand
affect lending in other markets connected to banks exposed to the shocks. Local credit demand
increases in response to disasters because residents need to re-build destroyed or damaged
physical capital. Local borrowers receive direct monetary support from the United States
Federal Emergency Management Association (FEMA), and they supplement these funds by
borrowing from banks. Banks themselves also are encouraged by their regulators to extend loans
to borrowers in areas that have been hit by natural disasters. In the first portion of our analysis,
we document that lending increases significantly during the months following disasters, with the
maximum increase occurring about 6 months after the shock.

1

Ashcraft (2006), Becker (2007), and Gilje (2012), for example, also show that the supply of local bank finance
affects investment.

2

To test how credit supply responds to exogenous demand increases, we focus on loan
originations in connected markets, those where banks lend before the disaster strikes but are not
directly affected by the natural disaster itself. Thus, identification assumes that loan demand in
(non-shocked) connected markets is unaffected by the natural disasters. To validate this
assumption, we report a placebo test whereby markets are randomly (and thus mostly falsely)
assigned as shocked. These tests reveal no change in lending to markets connected to the
placebo-shocked markets, validating the premise of our strategy.
To generate our empirical model, we build a panel dataset of loan originations at the
bank-county-month level. We use county to define the local credit market, and build the panel at
monthly level (rather than yearly) because the timing of the natural disasters is important.
Disasters strike in all months throughout the year but, as we will document, their effects on
demand dissipate to nearly zero within one year’s time. For lending, we use data on mortgage
originations reported to regulators under the Home Mortgage Disclosure Act (HMDA). These
are the only data that allow us to identify both the lending bank as well as the precise location of
the loan (based on the county of the property securing the mortgage). HMDA data are
sufficiently rich to allow us to estimate how changes in originations vary across different
segments of the mortgage market. For example, we find that originations decline most for
mortgages that can be easily securitized – those falling below the jumbo-loan threshold – which
suggests that banks protect their lending originations in the most profitable segments of the credit
markets. Despite this fall, rates of securitization of non-jumbo mortgages increases in connected
markets, especially for smaller banks.
We show that larger banks are better able to shield lending in their core markets from
demand shocks elsewhere relative to smaller banks. This difference reflects two advantages of
3

size. First, large banks have access to national debt and equity markets that can be used to
finance increased lending demand, while small banks rely heavily on deposits as their marginal
source of funds. Second, large banks lend in more markets where they have no branches, and
they tend to sell a much higher fraction of loans originated in these non-core markets, compared
to small banks. Our results support these advantages of size, as we find that large banks reduce
lending sharply in their non-core markets; in contrast, lending does not respond at all in the core
markets where large banks have branches. Small banks protect their core lending markets by
sharply increasing sales of their mortgage originations, but these sales are limited to the nonjumbo mortgage segment where the Government-Sponsored Enterprises (GSEs) – Fannie Mae
and Freddie Mac – are active buyers. Hence, small banks do reduce lending in core markets;
according to our estimates, these banks reduce lending in their core markets by 20-30 cents per
dollar of exposure to the natural disaster.
These differences point to an advantage of bank size that has not been emphasized much
in the literature, specifically that large banks can move capital into markets experiencing high
credit demand without needing to restrict credit in their core markets. Large, multi-market and
multi-product banks have numerous ways to move capital into areas with high demand. They
can finance the new lending by increasing loan sales and securitization, by reducing loan
origination in markets that can be served just as well by other lenders (e.g. markets where they
have no branches), and/or by raising external funds in capital markets. These advantages need to
be weighed against the costs of bank size related to subsidies from perceptions that they may be
‘too big to fail’ (Strahan, 2013).
A number of studies have used natural disasters to get exogenous variation in credit
conditions. Morse (2011) finds that poor residents fare better across a number of outcomes
4

following natural disasters in areas served by payday lenders. Chavaz (2014) shows that lenders
with concentrated exposure to markets hit by the massive hurricanes in 2005 increased lending
more than banks less concentrated in those areas. Consistent with this result, Cortés (2014) finds
that areas with a greater relative presence of local lenders recover faster after disasters.2 Our
approach exploits all natural disasters that occurred between 2001 and 2010, and measures the
effects of these well-defined events on actual lending growth. This approach allows us to build a
very rich dataset with many events (and thus many degrees of freedom); by using actual lending
changes in affected markets to measure the quantitative magnitude of these events, we can
include major hurricanes along with smaller and more localized shocks in a single empirical
framework.
Our study contributes to an emerging literature that tries to understand the role of banks
in integrating portions of local credit markets where arm’s length transactions (e.g.
securitization) are limited by information frictions. By allowing credit to flow between markets,
financial integration changes the effects of local credit-demand shocks. Ben-David, Palvia and
Spatt (2014) find that deposit rates paid increase when banks face strong external loan demand.
Loutskina and Strahan (2014) study the US housing boom of the 2000s and show that local
booms were made larger by capital inflows fostered by both securitization and branch banking.3
The increase in lending in booming areas such as Sun Belt states came at the expense of less2

Several other recent papers have focused on discrete and massive events, such as Hurricane Katrina. Lambert,
Noth and Schuwer (2012) find that banks attempt to accumulate capital following the shock. Massa and Zhang
(2013) use Katrina as an exogenous force leading insurance companies exposed to losses to fire sale bonds; they
find declines and later reversals in the prices of such sold bonds.

3

Previous research has also studied how financial integration through interstate banking reform affected economic
volatility and business cycle synchronization by allowing credit-supply shocks to be smoothed across geographies
(Morgan, Rime, and Strahan (2004); Demyanyk, Ostergaard, and Sorenson (2007)). Financial integration at the
global level have similarly been shown to help smooth shocks to the financial sector through cross-country risk
sharing (e.g. Peek and Rosengren, 2000; Bekaert, Harvey and Lundblad, 2005; Kalemli-Ozcan, Papaioannou, and
Peydró, 2010; Imai and Takarabe, 2011; and Cetorelli and Goldberg, 2012; Schnabl (2012).

5

booming areas. Consistent with this idea, Chakraborty, Goldstein, and Mackinlay (2013) find
that local business lending declined when banks reallocate capital toward areas with housing
booms, but that this result does not hold for large nationwide lenders. Our paper looks at the
same economic mechanism using a fully disaggregated approach and with a novel strategy to
identify exogenous credit demand shocks. Consistent with Chakraborty et al, we find that size
helps banks insulate their connected credit markets from demand shocks.
Our paper is most closely related to Gilje, Loutskina and Strahan (2014), who use the
HMDA mortgage originations data to study how financially integrated banks respond to
exogenous increases in funding availability from wealth inflows related to shale gas and oil
booms. In that study, banks receiving funding windfalls expand lending only in markets where
they have a branch presence. In this paper, we find that in response to higher demand for loans
in some markets, banks cut lending in connected markets most where they have no branch
presence.
What explains the asymmetry? Banks receiving positive liquidity windfalls optimally
expand the size of their balance sheet to take advantage of a lower cost of internal funds; such an
increase comes from both new loan originations as well as additional securities holdings
(Plosser, 2013). The increase in lending, however, only shows up in markets where banks have
an informational advantage based on the presence of a branch. Absent a source of market power
in lending, such as information or monitoring advantages from a local branch presence, funding
inflows are used to increase holdings of marketable securities rather than loans. In contrast,
banks that experience credit demand shocks that require additional funding reduce loans most in
markets where they possess little or no market power – markets without a branch presence. Thus,
banks appear to protect the rents that they can earn in core markets when they can. As noted
6

earlier, small banks have fewer funding options than larger ones and, as we show, cut lending
more in their core markets.4
2. DATA & EMPIRICAL METHODS
2.1 Data
The Spatial Hazard Events and Losses Database for the United States (SHELDUS) is a
county-level hazard data set covering the U.S., with different natural hazard event-types such as
thunderstorms, hurricanes, floods, wildfires, and tornados. For each event, the database includes
the beginning date, location (county and state), property losses, crop losses, injuries, and
fatalities that affected each county. The data were derived from several existing national data
sources such as National Climatic Data Center’s monthly storm data publications. Our sample
starts with all natural disasters reported in SHELDUS that occurred in the US states between
2001 and 2010 and includes those in which the Governor declared a ‘state of emergency’ with a
formal request for Federal Emergency Management Agency (FEMA) funds to respond to the
disaster. Thus, we include only relatively large disasters.
Table 1 reports summary statistics on the number of affected counties, total property
damages, and the distribution of property damages across eight types of disasters. Overall there
are 5,501 counties affected by the disasters (about 500 per year). Hurricanes, while relatively
rare, each affect a large number of counties per event due to their massive scale, so we have
more than 1,500 counties affected by them. Severe storms affect even more counties (over 3,000
in total) due to their high frequency, even though each one is typically limited in scope. All of
the disasters in our sample are severe because since a ‘state of emergency’ had to have been
4

Berrospide et al (2013) find that banks also protect their core markets from declines in lending stemming from
lender distress during the housing collapse in 2007-2009.

7

declared, but that severity varies substantially by type. Most of the disasters mete our relatively
small losses at the median, but all types mete out significant damages in the tails of the
distribution. For example, tornado losses exceeded $160 million at the 99th percentile; hurricane
losses exceeded $1.3 billion in tail events; even severe blizzard losses exceeded $6 million at the
top end of the distribution. As we describe in detail below, we construct our variables to account
for the severity of each event.
In our core models, we measure lending outcomes at the bank-county-month level,
focusing on data on mortgage originations collected under the Home Mortgage Disclosure Act
(HMDA). The annual publicly available version of the HMDA data does not include the exact
date for each loan-application record, but we have access to the confidential version of these
data, which do allow us to measure mortgage originations at monthly frequency. Information on
timing is important in our setting because, as we show below, the effects of disasters on credit
dwindle to zero by 12 months post-shock. Whether a lender is covered in HMDA depends on its
size, the extent of its activity in a Central Business Statistical Area, and the weight of residential
mortgage lending in its portfolio. That said, the bulk of residential mortgage lending activity is
likely to be reported.5 We map the HMDA data into bank asset size and branch location data
from June of the prior year using the FDIC’s Summary of Deposits data.
We drop large banks - those with total assets above $2 billion – from our study, leaving
us with a sample of small and medium-sized banks. While we think that capital flows within

5

Any depository institution with a home office or branch in a CBSA must report HMDA data if it has made a home
purchase loan on a one-to-four unit dwelling or has refinanced a home purchase loan and if it has assets above $30
million. Any non-depository institution with at least ten percent of its loan portfolio composed of home purchase
loans must also report HMDA data if it has assets exceeding $10 million. Consequently, HMDA data does not
capture lending activity of small or rural originators. U.S. Census shows that about 83 percent of the population
lived in metropolitan areas over our sample period.

8

large banking organizations are important, our preliminary empirical analysis suggests that the
shocks driving credit demand variation are just too small to have a meaningful impact on the
largest institutions. For example, even a shock as large as Hurricane Katrina affected only about
5% of the 2,777 counties in which Bank of America actively supplied mortgage credit in 2005.
Most of the natural disasters in our data are, of course, much smaller and more localized than
Katrina, and thus would have minimal effects on the credit demand faced by very large banks.
The HMDA data include loan size, whether or not a loan was approved, as well as some
information on borrower characteristics. Loans above a certain size cutoff may not be sold to
one of the Government-Sponsored Housing Enterprises, Fannie Mae and Freddie Mac (the
GSEs). Jumbo mortgages are thus less liquid than non-jumbos (Loutskina & Strahan, 2009), so
we will disaggregate our results based on this size cutoff in some of our tests. HMDA reports
both the identity of the lender as well as the location of the property down to the census-tract
level. These are the only comprehensive data on lending by US banks that allow us to locate
borrowers geographically.6 HMDA also contains information on the purpose of the loan
(mortgage purchase loans, home-equity loans, and mortgage re-financings). We include only
mortgages for home purchase in our tests. HMDA also flags whether the lender expects to sell
or securitize the loan within one year of origination. We use this flag to test whether loans easier
to finance in securitization markets respond differentially to the local credit demand shocks.
2.2 Natural Disasters as a Shock to Credit Demand
We use the FEMA-disaster subset of the SHELDUS data to measure exogenous changes
in credit demand at the local level. Demand increases after disasters because affected residents
6

Other loans types of loans, particularly those to small businesses who depend on banks, would be interesting to
study. Detailed and comprehensive data by lender and location, however, are scarce for the small business sector.

9

need to rebuild damaged homes and businesses. Some of the funds for rebuilding come from
FEMA directly, and affected individuals supplement these funds by borrowings from banks. In
fact, banks themselves are encouraged to increase credit availability after many of these shocks
by their regulators. Following the flooding in Colorado in 2013, for instance, the Federal
Deposit Insurance Corporation (FDIC) issued a Financial Institutions Letter to local lenders
(FIL-39-2013) with the following language: “The FDIC has announced a series of steps intended
to provide regulatory relief to financial institutions and facilitate recovery in areas of Colorado
affected by severe storms, flooding, landslides, and mudslides.” And, “Extending repayment
terms, restructuring existing loans, or easing terms for new loans, if done in a manner consistent
with sound banking practices, can contribute to the health of the local community and serve the
long-term interests of the lending institution.”
To validate the basic premise of our identification strategy, we first test whether lending
is abnormally high in the months immediately following natural disasters. We do so by
constructing a panel dataset at the county-month level of (the log of) total mortgage applications
(including applications to all lenders, not just those in our sample), and regress this variable on
county and month fixed effects, plus a series of event-time indicator variables defined around the
date of each natural disaster, as follows:
Log Mortgage Originationsj,t = αj + γt + Σβk Dkj,t + εj,t ,

(1)

where j indexes counties (αj are county-level fixed effects), and t indexes months (γt are time
effects). Event-time indicators (Dkj,t ) run from k = -3 (3 months before the disaster) to k = +12
(12 months after the disaster), where k = 0 represent the month in which the shock itself occurs.

10

Figure 1 reports the β coefficients from our estimate of (1), along with standard error
boundaries representing the 95th percentile confidence interval around them. These coefficients
measure abnormal mortgage originations, relative to each county’s long-run average (absorbed
by the αj) and relative to the time-average across all counties (absorbed by the γt). Figure 1
shows no abnormally high or low levels of lending before the disaster (consistent with the
disaster being exogenous and unexpected). The F-test on the pre-shock indicators equals 0.39
with a p-value of 0.76. Abnormally high levels of lending do occur after the disasters, starting in
month +2, consistent with an increase in loan demand due to the shock. The F-test on the postshock indicators equals 3.65 with a p-value of 0.0006. The increase in lending peaks about 6
months after the shock (about 3% above normal), and then dissipates by the end of 12 months.
The preliminary results in Figure 1 key off lending in the mortgage market, which is
likely not the only (or even the main) lending market affected by natural disasters. For example,
construction loans are likely to be spurred substantially by the need of local residents to rebuild.
Consistent with the idea that overall credit demand rises, Cortés (2014) uses Call Report data to
show that small banks with at least 65% of their branches in one market increase total lending by
about 25% during the year following a local natural disaster, and that most of that increase
occurs in the two quarters following the shock.7
2.3 Modeling how Demand Shocks Affect Lending in Connected Markets
To study capital movements within banks, we build a panel dataset at the bank-countymonth level, using the HMDA data on mortgage originations from 2001 to 2010. For each bankmonth, we include all of the counties in which that bank originated some mortgages in the prior
7

Call Report data do not report information on borrower location; hence, it is not useful for understanding lending
by multi-market banks across their various markets, as we focus on below.

11

year. These counties are assumed to be the relevant lending markets for each bank. For
example, if a bank originated mortgages in 25 counties last year, that bank would generate 300
observations this year (=12 months times 25 counties). We then flag each county in the month in
which that county experienced a natural disaster, and leave that flag on during the next 12
months. Changes in lending during these 12 months are assumed to stem from extra credit
demand due to the shock (recall Figure 1). We drop these ‘shocked’ county-months from our
bank-county-month dataset because our aim is to test how the shock affects lending in connected
markets.8 The incremental lending by each bank in the shocked county-months provides a proxy
for the higher demand experienced by these banks as a consequence of the natural disaster.
Since banks operate across different numbers of connected (non-shocked) markets, we parcel out
the increase equally across each of these markets. Analytically,
Shocki,t = ΔLending-in-shocked-countiesi,t / Ni,t ,

(2)

where i represents banks and t represents months. The variable ΔLending-in-shocked-countiesi,t
equals the change in the total dollar-value of mortgage loans between month t and month t-1
originated by bank i, summed across all markets in which bank i operates that are flagged as
shocked in month t; Ni,t equals the number of non-shocked markets connected to bank i in month
t. Notice that the shock varies at the bank-month level (as opposed to the bank-county-month
level).
We estimate the effect of each bank’s additional lending from the demand increase in the
shocked areas on its lending originations in connected (non-shocked) markets, as follows:

8

We also drop these shocked counties for an additional 12 months to be sure that credit demand there has returned
to normal.

12

ΔLendingi,j,t / Total Lendingi,t = αi,j + γj,t + Σ βkShocki,t-k / Total Lendingi,t + εi,j,t , (3)
where j indexes counties, i indexes banks, t indexes months, and k indexes lags of the exposure
variable (we include 12 lags).9 County-month effects (γj,t) sweep out potentially confounding
factors affecting all lenders in a given county-month (such as unobserved local credit demand
shocks, business cycle effects, trends, etc.). We also remove bank-county effects (αi,j), although
we introduce interaction effects between bank characteristics and the shocks in some models.
We divide both dependent and the key explanatory variables by each bank’s total lending in
month t as a normalization that will help reduce heteroskedasticity.10 Note that banks operating
in just one market play no direct role in estimating the βk coefficients, since their exposure to
natural disasters in non-shocked markets would always equal zero. We leave them in the model,
however, because they help pin down the γj,t and thus improve the model’s power to sweep out
potentially confounding credit-demand effects.
The magnitude of shocks, which differ widely depending on the severity of disasters, is
captured implicitly because we measure the total change in lending experienced by a bank in all
of its shocked areas. For example, a string of tornados hit 14 counties in Ohio in August 2003,
and on average banks lent $15 million more per month in the year following the disaster in the
affected counties than in the six months prior to the shock. Lending changes will be large
following large shocks (e.g. Katrina) and small following smaller ones (e.g. severe storms,
blizzards, etc.).

9

Abnormal loan volume following natural disasters declines to zero by 12 months out, as shown in Figure 1. But
we have also estimated equation (3) with 18 and 24 lags and find that these additional lags are small and not
statistically significant.
10

Both the change in lending and the explanatory shock are bounded between -1 and 1 when scaled by total lending
by design in order to measure succinctly the resulting contraction in lending in dollars in a response to a shock.

13

We also include the log of bank assets as an additional time-varying, bank-level control
variable (time invariant bank-county characteristics, such as the exact distribution of its
branches, get absorbed by the αi,j).11 Since the regression is built from dollar-changes in lending
(normalized) parceled evenly across markets, the sum of the β coefficients from equation (3) can
be interpreted as the total effect per dollar of increased lending in the shocked market on lending
in the bank’s connected, non-shocked markets. Thus, we expect the sum of these coefficients to
lie between zero and -1. Since the key variables of interest – each bank’s lags of exposure to the
demand shocks – do not vary across counties, we cluster by bank in building standard errors.12
Table 2 reports summary statistics for the panel data used to estimate our regressions.
There are 7,336,224 bank-month-county observations in the main sample from 2002-2010. We
use disasters from 2001 to identify the lags for 2002 going back 12 months, so 2001 does not
appear in the regression. The mean for the dependent variable equals 0.0356. The mean of the
key explanatory variable Shocki,t equals 0.0026 (compared to just 0.0008 for the placebo
specification). Recall that Shocki,t has a zero value for many observations: all bank-countymonth observations in which the bank did not have any exposure to a market experiencing a
natural disasters over the past year. For non-zero values, Shocki,t averages 0.0069. Since the
average bank has 18 non-shocked markets at the time of a disaster, the average growth in lending
in the shocked markets equals about 12.4% (0.0069 * 18). While not displayed in Table 2, there
are 6,414 unique banks in the sample, and the median number of branches per bank is 5.

11

Our main results are not sensitive to whether or not we include the bank-size measure; most of the effects of bank
characteristics are absorbed by the bank*county fixed effects.

12

We have also tried clustering by state and estimate standard errors smaller than those reported here.

14

3. RESULTS
First, we report our baseline model for all bank-county-year observations, and compare
those results to the placebo test (Table 3). Second, we separate the sample to explore how
variation across market types (core markets, those with branches v. non-core markets, those
without) and bank types (larger banks v. smaller ones) affects responses to shocks (Tables 4 &
5). Third, we go back to the pooled approach and disaggregate changes in lending (the
dependent variable) based on mortgage size (jumbo v. non-jumbo) and whether or not a loan is
retained or sold by the originating bank (Tables 6 & 7). Last, we combine the second and third
approaches (Tables 8-10).
3.1 Baseline Model & Placebo Test
Table 3, column 1 reports the simplest model with all bank-county-years included. We
report the coefficients on the 12 lags of exposure to the shock. These shocks are highly
persistent by construction because we allow a given county’s exposure to a disaster to last for 12
consecutive months. Individual coefficients are sometimes hard to estimate precisely (especially
in later tests where we introduce interactions) due to multi-collinearity across the 12 lags. Thus,
we focus most of our attention on the long-run effects (the sum of the coefficients), rather than
on the individual lagged effects and the implied dynamics of those coefficients. The sum
estimates the total impact on lending in connected markets per dollar of increased lending in
shocked markets.
We find that lending falls by about 50 cents per dollar of additional lending stimulated by
the shock exposure (i.e., the sum of the coefficients on the twelve lags = -0.504). The effect is
large economically, but is also statistically significantly smaller in magnitude than -1, meaning
15

that banks increase their overall lending in response to natural disasters. (A coefficient sum equal
to -1 would imply that all of the extra lending in the shocked localities displaces lending in other
markets.) Thus, banks are able to protect credit supply partially, but not fully, by selling
securities, increasing loan sales and/or raising additional debt and equity funds.
Column 2 of Table 3 reports the results from the placebo test, which uses the exact same
structure and data, but assigns markets as ‘shocked’ randomly. In setting up this test, we
preserve the number and temporal distribution of the local natural disasters, but we assign them
randomly across markets. We find no significant correlation between the (mostly falsely
assigned) placebo exposure measures and actual lending in connected markets; the sum of the
coefficients on the 12 lags is small and not significant, as are each of the individual coefficients
on the 12 lags.
3.2 Variation across market-types and bank-types
Next, we test how variation in credit supply depends on market characteristics and bank
characteristics. For these tests, we define core markets as those counties where a bank lent in the
prior year with a branch presence; non-core markets are defined as counties where a bank lent in
the prior year but without a branch presence. We divide lenders based on the number of counties
(markets) in which they lent in the prior year, where ‘small’ banks are those active in fewer than
15 markets (the in-sample median) and ‘large’ banks are those active in more than 15 markets.13
Table 4 compares core v. non-core markets by introducing the Branch indicator and its
interaction with the disaster exposure measures. This model allows the amount by which lending
falls with exposure to shocks to vary across market types. The effect on non-core markets equals
13

Results are similar if we divide the sample of banks by total assets rather than by number of markets.

16

the sum across the first 12 lags, while the effect on core-market lending equals the sum across
these 12 lags plus the additional 12 interaction terms. This model shows that banks protect
lending in core areas. Lending falls by about 55 cents per dollar in non-core areas but just 25
cents per dollar in core areas. Both of these coefficients sums are statistically significantly
greater than zero, and they are also significantly different from each other. The direct effect of
the Branch indicator, however, is not statistically significant because most of its effects are
wiped out by the bank*county fixed effect.14
Table 5 puts market type and bank size together by estimating the models of Table 4 after
splitting the sample: Panel A reports results for small banks and Panel B for large ones. There
are twice as many small banks as there are large – 5,457 compared to 2,604 – but the observation
number in the regressions is much larger in Panel B because large banks operate in more markets
(by definition). These results highlight a key advantage of size. Because large banks operate in
many non-core areas, they have more freedom to reduce lending in those areas to accommodate
demand shocks; on average, non-core markets represent 17% of all markets in which large banks
lend. As Panel B shows, in non-core markets large banks reduce lending sharply, whereas we
find no significant decline on lending in core markets. In contrast, small banks exhibit no such
difference, in large part because they tend to lend almost exclusively in markets where they have
branches; in contrast to large banks, non-core markets represent just 5% of the markets in which
small banks lend. Thus, it makes sense that the branch interaction effects are not statistically
significant for this sample (insufficient variation).
3.3 Variation across loan-types
14

We hesitate to over interpret the time series dynamics implied by our regressors, but the interactive effects in
Table 4 suggest that lending falls sharply immediately after disasters and then rebounds thereafter. The initial sharp
drop may reflect capacity constraints in labor markets if the shocked banks re-deploy bank lending officers to the
shocked markets from their core markets.

17

As described in Loutskina and Strahan (2009), the mortgage market has been segmented
by the activities of the GSEs – Fannie Mae and Freddie Mac. The GSEs enhance liquidity by
buying mortgages directly from lenders and also by selling credit protection that allows such
mortgages to be securitized easily by the originator. Yet the GSEs operate under a special
charter limiting the size (and credit risk) of mortgages that they may purchase or help securitize.
These limitations were designed to ensure that the GSEs meet the legislative goal of promoting
access to mortgage credit for low- and moderate-income households. The GSEs may thus only
purchase non-jumbo mortgages, those below a given size threshold. Until the Financial Crisis,
the jumbo-loan limit increased each year by the percentage change in the national average of
single-family housing prices, based on a survey of major lenders by the Federal Housing Finance
Board. For example, in 2006 the jumbo-loan limit was $417,000 for loans secured by singlefamily homes. (The limit is 50% higher in Alaska and Hawaii so we exclude them from the
sample in the tests on loan type.) After 2007, the practice of tying the jumbo-loan cutoff to
nationwide house-price changes was abandoned in an effort to subsidize mortgage finance and
slow the decline in house prices. For example, rather than reduce the cutoff as housing prices
fell, they were actually maintained or increased. Moreover, after this time the jumbo-loan cutoff
was changed to reflect the level of average prices across markets. Thus, the importance of GSEs
in mortgage finance increased after the Crisis.
With the actions of the GSEs, the non-jumbo mortgage markets tend to be both more
competitive and more liquid than the jumbo segment. Competition tends to reduce the
profitability of the non-jumbo segment, whereas liquidity tends to reduce the extent to which

18

banks need to finance these mortgages locally.15 For example, banks facing increased credit
demand elsewhere (due to natural disasters or other reasons) may respond by increasing the
extent to which non-jumbo mortgages are sold or securitized.
Tables 6 and 7 test how credit supply responds to the natural disaster exposure for
different mortgage-market segments (jumbo v. non-jumbo), and whether or not lenders expect to
sell or retain the mortgage. In Table 6, we split the dependent variable (ΔLendingi,j,t / Total
Lendingi,t) into two pieces that sum to the original one:
ΔNon-Jumbo Lendingi,j,t / Total Lendingi,t + ΔJumbo Lendingi,j,t / Total Lendingi,t =
ΔLendingi,j,t / Total Lendingi,t.

(4)

Thus, coefficients ‘add up’ across columns 1 and 2 of Table 6 to those reported in Table 3.16
Table 7 further sub-divides the dependent variable into four components that add to the total
change in lending:
ΔNon-Jumbo Sold Lendingi,j,t / Total Lendingi,t + ΔNon-Jumbo Retained Lendingi,j,t /
Total Lendingi,t + ΔJumbo Sold Lendingi,j,t / Total Lendingi,t + ΔJumbo Retained
Lendingi,j,t / Total Lendingi,t = ΔLendingi,j,t / Total Lendingi,t.

(5)

As shown in Table 6, non-jumbo lending declines much more than jumbo lending. This
difference reflects two forces both leading to the same outcome: the non-jumbo segment is

15

Scharfstein and Sunderam (2013) show that markets with greater lender concentration are less competitive,
leading to an increase in the difference between the price of mortgages to borrowers relative to the financing costs in
the mortgage-backed securities market.

16

This adding up would hold exactly if the samples were identical between Table 3 and Table 6. They are not
because we lose some observations when we disaggregate the data into the two segments by dropping Alaska and
Hawaii, where the jumbo cutoff is 50% higher than in the contiguous states.

19

quantitatively larger, so there are more dollars of lending that can be siphoned off to other
markets; and, the non-jumbo segment is more competitive, so reducing a given dollar of credit in
that segment is less costly to banks in terms of foregone profits.
Table 7, however, shows that more than 100% of the decline in lending in the non-jumbo
segment comes from declines in retained mortgages, whereas mortgages sold actual increases.
Thus, banks use securitization to substitute for on-balance sheet finance required to lend in shock
markets, thus mitigating (partially) the decline in loan originations in connected markets. This
substitution is much more feasible in the non-jumbo segment because of the actions of the GSEs,
which grease the wheels of the securitization process.17 In the jumbo segment, the point estimate
for sold loans is also positive, but it is small and not statistically significant.
3.4 Variation across market-, bank-, and loan-types
In our last three Tables (8-10), we disaggregate results by bank size, loan type and market
type. Table 8 reports results for changes in non-jumbo and jumbo lending (retained and sold) for
small banks; Table 9 reports similar models for large banks. In Table 10, we add the Branch
indicator and its interactions with disaster exposure to the large-bank models to account for
market type. (We leave out the small-bank analogue to Table 10 because, as noted earlier, small
banks lend almost exclusively in core markets; as such, the interaction terms with Branch
indicator were not significant for the small-bank sample (recall Table 5, Panel A)).
For small banks (Table 8), all of the declines in mortgage originations occur in nonjumbo, retained mortgages. The coefficients suggest that lending falls in this segment by about
57 cents per dollar of increased lending in shocked markets. In contrast, the point estimate for
17

The frequency with which loans are sold falls by about 25 percentage points, comparing non-jumbo with jumbo
mortgages. The drop happens discontinuously around the cutoff (Loutskina and Strahan, 2011).

20

the sold loan originations is large and positive, offsetting about 2/3s of the decline in retained
mortgage originations. So, small banks substitute some of the funding needed to supply credit to
shocked markets by selling more of their loan originations. This finding supports Loutskina
(2011), who shows that access to securitization markets enhances loan liquidity and makes banks
less sensitive to changes in funding costs from monetary policy actions. The results here imply
that securitization makes banks less sensitive to changes in loan demand in external markets. In
the jumbo segment, loan sales and securitization are much more costly; hence this mechanism is
not available.
Patterns for large banks differ from those of smaller ones. For them, overall lending
declines for both sold and retained loans in the non-jumbo segment, and also for retained
mortgages in the jumbo segment (Table 9). However, when we introduce the Branch indicator
and its interactions, which allow us to compare adjustments between core and non-core markets,
we see all of the declines in lending happen in large banks’ non-core markets. In their core
markets, we find no significant change in mortgage originations (Table 10).
4. CONCLUSIONS
In this paper we trace out movements of capital within multi-market banks. Credit
demand increases in response to local shocks created by exposure to natural disasters. Banks
respond by increasing credit in those areas, and by taking credit away from other markets in
which they have lent. But banks mitigate the potential reductions of credit to connected markets
using the loan sales/securitization markets rather than holding originated loans, as we find that
small banks increase their use of these markets to mitigate declines in overall mortgage
origination. This mechanism only partially shields credit supplied by small banks in their core
markets connected to the shocked markets. Larger banks, in contrast, reduce credit sharply in
21

their non-core markets, obviating the need to reduce credit in areas where they have local market
expertise.

REFERENCES
Agrawal, S., Hauswald, R., 2010, “Distance and Private Information in Lending,” Review of Financial
Studies 23, 2757-2788.
Ashcraft, A., 2006, “New Evidence on the Lending Channel,” Journal of Money, Credit, and Banking 38,
751-776.
Bekaert, Geert, Campbell R. Harvey and Christian Lundblad, 2005 “Does Financial Liberalization Spur
Growth,” Journal of Financial Economics 77, 3-55.
Becker, Bo, 2007 “Geographical Segmentation of US Capital Markets”, Journal of Financial Economics
85(1), 151-178.
Berrospide J. M., Black, L. P., Keeton, W. R., 2013, “The Cross-Market Spillover of Economic Shocks
through Multi-Market Banks,” Working Paper.
Ben-David, I., Palvia, A, Spatt, C., 2013, "Internal Capital Markets and Deposit Rates," Working Paper
Berger, A. N., Miller, N. H., Petersen, M. A., Rajan, R. G., Stein, J. C., 2005, “Does Function Follow
Organizational For? Evidence from the Lending Practices of Large and Small Banks,” Journal of
Financial Economics 76, 237-269.
Cetorelli, N., Goldberg, L., 2012, “Bank Globalization and Monetary Policy,” Journal of Finance.
Chavaz, M., 2014, “Riders of the Storm: Economic Shock & Bank Lending in a Natural Experiment”
Working Paper.
Chakraborty, I., Goldstein, I., Mackinlay, A., 2013, “Do Asset Price Bubbles have Negative Real
Effects?” Working Paper.
Cortés, K. R., 2012 “Did Local Lenders Forecast the Bust? Evidence from the Real Estate Market,”
Working Paper.
22

Cortés, K. R., 2014, “Rebuilding after Disaster Strikes: How Local Lenders Aid in the Recovery,”
Working Paper.
Degryse, H., Ongena, S., 2005, “Distance, Lending Relationships, and Competition,” Journal of Finance
60, 231-266.
Ergungor, O. E., 2010, “Bank Branch Presence and Access to Credit in Low-to-Moderate Income
Households,” Journal of Money, Credit and Banking 42(7).
Gilje, E. P., 2012, “Does Local Access To Finance Matter? Evidence from U.S. Oil and Natural Gas
Shale Booms,” Working Paper.
Gilje, E. P., Loutskina, E. and Strahan, P. E., 2014, “Exporting Liquidity: Branch Banking and Financial
Integration,” Working Paper.
Imai, M.,Takarabe, S. 2011, "Bank Integration and Transmission of Financial Shocks: Evidence from J
Japan." American Economic Journal: Macroeconomics, 3(1): 155-83.
Kalemili-Ozcan, S., Papaioannou, E., Peydró, J., 2013, “Financial Regulation, Financial Globalization,
and the Synchronization of Economic Activity,” Journal of Finance 68, 1179-1228.
Lambert, C., F. Noth, and U. Schuwer, 2012, “How do banks react to increased asset risk? Evidence from
Hurricane Katrina” (AEA Meetings Paper).
Loutskina, E., 2011, “The Role of Securitization in Bank Liquidity and Funding Management,” Journal
of Financial Economics 100, 663-684.
Loutskina, E, Strahan, P.E., 2009, “Securitization and the Declining Impact of Bank Finance on Loan
Supply: Evidence from Mortgage Acceptance Rates,” Journal of Finance 64, 861-889.
Loutskina, E., Strahan, P. E., 2011, “Informed and Uninformed Investment in Housing: The Downside of
Diversification,” Review of Financial Studies 24, 1447-1480.
Loutskina, E., Strahan, P. E., forthcoming, “Financial Integration, Housing and Economic Volatility,”
Journal of Financial Economics.
Massa, M., Zhang, L., 2013, The spillover effect of hurricane Katrina on corporate bonds and the choice
between bank and bond financing,” Working Paper.
Morgan, D. P., Rime, B., Strahan, P. E., 2004, “Bank Integration and State Business Cycles,” Quarterly
Journal of Economics 119, 1555-1585.
Morse, Adair, 2011, “Payday Lenders: Heroes or Villains?” Journal of Financial Economics 102(1), 2844.
Peek, J., Rosengren, E., 1997, “The International Transmission of Financial Shocks: The Case of Japan,”
American Economic Review 87, 495-505.
23

Plosser, M., 2013, “Bank Heterogeneity and Capital Allocation: Evidence from 'Fracking' Shocks,”
Working Paper.
Scharfstein, D., Sunderam, A., 2013, “Concentration in Mortgage Lending, Refinancing Activity, and
Mortgage Rates,” Working Paper.
Schnabl, P., 2012, “The International Transmission of Bank Liquidity Shocks: Evidence from an
Emerging Market,” Journal of Finance 67, 897-932.
Strahan, Philip E., 2013, “Too Big to Fail: Causes, Consequences and Policy Responses,” Annual Review
of Financial Economics.

24

Log of Mortgage Purchase Loan Originations

25

0.08
0.06
0.04
0.02
0
‐0.02
‐0.04
‐0.06

Fig. 1: Log of Mortgage Originations around
Natural Disasters
(With 95% Confidence Intervals)

Table 1: Property Damage from Natural Disasters
This table reports data on propery losses from natural disasters taken from the Spatial Hazard Events and Losses Database for the United States. These
data are at county-level. The sample starts with all natural disasters reported in SHELDUS that occurred in the US states between 2001 and 2010 and
includes those in which the Governor declared a ‘state of emergency’ with a formal request for Federal Emergency Management Agency (FEMA) funds
to respond to the disaster.
Property Damage Distribution
Total Property Across
25th
75th
95th
99th
Natural Disaster
Number of Affected Counties all Counties (Billions) Percentile
Median
Percentile
Percentile
Percentile
----------------(Millions of $s)------------------------------------Coastal
68
4.82
0.1
1.0
3.2
340.0
340.0
Wildfire
280
3.43
0.0
0.1
0.4
24.0
593.0
Earthquake
16
0.08
0.0
0.0
10.6
14.6
14.6
Flooding
108
0.30
0.0
0.5
2.0
15.0
22.1
Hurricane
1,545
121
0.0
0.3
5.0
250.0
1,330.0
Severe Storm (Ice, Hail)
3,169
15
0.0
0.1
0.9
16.0
75.0
Blizzard
252
0.14
0.0
0.1
0.4
3.6
6.2
Tornado
63
0.27
0.0
0.1
0.6
15.2
130.0
All disaster types

5,501

146

0.0

0.1

1.4

44.6

510.0

Table 2: Summary Statistics Regression Variables
This table reports summary statistics for the change in monthly mortgage originations and disaster exposure used in our baseline regression
models. The data are measured at the bank-county-month level, including all counties where a bank has lent in the prior year. There are
14,419 distinct banks in the sample, which spans the years 2002 to 2010. Disaster Exposure equals the change in the total dollar-value of
mortgage loans between month t and month t-1 originated by bank i, summed across all markets in which bank i operates that are flagged as
having been shocked by a natural disaster in month t; we divide this by the number of non-shocked markets connected to bank i in month t.
Placebo exposure is defined similarly to disaster exposure, with the difference being that shocked markets are chosen randomly. Both the
change in mortgage lending and disaster exposure are normalized by each banks total mortgage originations across all of its markets.

Change in Monthly Mortgage Originations by Total Lending
Disaster Exposure
Placebo Exposure
Size (Log of Assets)
Assets (In Thousands $s)

Observations
7,336,224
7,336,224
7,336,224
7,336,224
7,336,224

Mean
0.0356
0.0026
0.0008
12.88
601,221

Std. Dev.
0.24
0.03
0.04
0.99
515,445

Table 3: The effect of credit demand shocks on connected markets
This table reports regressions of the change mortgage originations for bank i/county j/month t on the change in lending in
counties hit by natural disasters. The models include County x Month fixed effects (to absorb demand shocks) and Bank x
County fixed effects (to absorb unobserved heterogeneity), with standard errors clustered by bank. The sample includes all
banks with assets above $2 billion. A county is included if the bank originated any mortgages in the prior year.
Dependent Variable : Change in Mortgage Originationsi,j,t / Mortgage originationsi,t

Disaster Exposurei,t-1
Disaster Exposurei,t-2
Disaster Exposurei,t-3
Disaster Exposurei,t-4
Disaster Exposurei,t-5
Disaster Exposurei,t-6
Disaster Exposurei,t-7
Disaster Exposurei,t-8
Disaster Exposurei,t-9
Disaster Exposurei,t-10
Disaster Exposurei,t-11
Disaster Exposurei,t-12
Log of Bank Assets
Coefficient Sum
F( 1,6765487)
P-Value
County by Time FE
Bank by County FE
Number of clusters (banks)
Observations
R-squared
*** p<0.01, ** p<0.05, * p<0.1

Disaster Exposure Based on Actually
Shocked Markets
-0.0366***
(0.004)
-0.0373***
(0.004)
-0.0460***
(0.005)
-0.0506***
(0.005)
-0.0598***
(0.006)
-0.0564***
(0.006)
-0.0767***
(0.007)
-0.0519***
(0.007)
-0.0327***
(0.006)
-0.0249***
(0.006)
-0.0200***
(0.006)
-0.0107**
(0.005)
-0.0202***
(0.003)
-0.504
303.7
0.0000

Placebo: Randomly assigned 'shocked'
markets
0.00453
(0.006)
-0.00416
(0.007)
-0.00444
(0.006)
0.0011
(0.007)
0.00749
(0.006)
-0.00764
(0.009)
-0.00384
(0.006)
-0.00449
(0.009)
0.00125
(0.005)
-0.00845
(0.010)
-0.001
(0.006)
0.000228
(0.006)
-0.0203***
(0.003)
-0.019
0.44
0.506

Yes
Yes
6,414
7,335,675
0.286

Yes
Yes
6,414
7,335,675
0.286

Table 4: The effect of credit demand shocks on connected markets, Core v. Non-Core Markets
This table reports the regression of the change mortgage originations for bank i/county j/month t on the change in lending in counties
hit by natural disasters, and allows the effects to vary between core markets (those with branches) and non-core markets (those
without). The models include County x Month fixed effects (to absorb demand shocks) and Bank x County fixed effects (to absorb
unobserved heterogeneity), with standard errors clustered by bank. The sample includes all banks with assets above $2 billion. A
county is included if the bank originated any mortgages in the prior year.
Dependent Variable : Change in Mortgage Originationsi,j,t / Mortgage originationsi,t
Disaster Exposurei,t-1
Disaster Exposurei,t-2
Disaster Exposurei,t-3
Disaster Exposurei,t-4
Disaster Exposurei,t-5
Disaster Exposurei,t-6
Disaster Exposurei,t-7
Disaster Exposurei,t-8
Disaster Exposurei,t-9
Disaster Exposurei,t-10
Disaster Exposurei,t-11
Disaster Exposurei,t-12
Log of Bank Assets

Coefficient Sum
F
P-Value
County by Time FE
Bank by County FE
Number of clusters (banks)
Observations
R-squared
*** p<0.01, ** p<0.05, * p<0.1

-0.0264***
(0.003)
-0.0335***
(0.004)
-0.0372***
(0.004)
-0.0460***
(0.005)
-0.0608***
(0.005)
-0.0628***
(0.006)
-0.0841***
(0.007)
-0.0663***
(0.006)
-0.0450***
(0.006)
-0.0402***
(0.006)
-0.0314***
(0.005)
-0.0154***
(0.005)
-0.0203***
(0.003)
Markets w/o branches
-0.549
366.03
0.0000

Branch * Disaster Exposurei,t-1
Branch * Disaster Exposurei,t-2
Branch * Disaster Exposurei,t-3
Branch * Disaster Exposurei,t-4
Branch * Disaster Exposurei,t-5
Branch * Disaster Exposurei,t-6
Branch * Disaster Exposurei,t-7
Branch * Disaster Exposurei,t-8
Branch * Disaster Exposurei,t-9
Branch * Disaster Exposurei,t-10
Branch * Disaster Exposurei,t-11
Branch * Disaster Exposurei,t-12
Branchi,j,t-1

Yes
Yes
6,414
7,335,675
0.286

-0.0718***
(0.017)
-0.0336*
(0.018)
-0.0650***
(0.020)
-0.0383*
(0.021)
0.00213
(0.022)
0.0395
(0.024)
0.0521**
(0.026)
0.0991***
(0.026)
0.0892***
(0.026)
0.109***
(0.025)
0.0824***
(0.024)
0.0348
(0.024)
0.00112
(0.002)
Markets with Branches
-0.250
7.52
0.0061

Table 5: The effect of credit demand shocks on connected markets, Core v. Non-Core Markets: Large v. Small Banks
This table reports the regression of the change mortgage originations for bank i/county j/month t on the change in lending in counties hit by natural disasters, and allows the effects to vary between core market
(those with branches) and non-core markets (those without). The models include County x Month fixed effects (to absorb demand shocks) and Bank x County fixed effects (to absorb unobserved
heterogeneity), with standard errors clustered by bank. The sample includes all banks with assets above $2 billion. A county is included if the bank originated any mortgages in the prior year.
Dependent Variable : Change in Mortgage Originationsi,j,t / Mortgage originationsi,t
Panel A: Small Banks (15 or Fewer Lending Markets)
Panel B: Large Banks (More than 15 Lending Markets)
Disaster Exposurei,t-1
Branch * Disaster Exposurei,t-1
Disaster Exposurei,t-1
Branch * Disaster Exposurei,t-1
-0.0151***
-0.0581***
-0.0624***
-0.165***
(0.004)
(0.020)
(0.008)
(0.042)
Branch * Disaster Exposurei,t-2
Disaster Exposurei,t-2
Branch * Disaster Exposurei,t-2
Disaster Exposurei,t-2
-0.0160***
-0.0321
-0.0628***
-0.0223
(0.005)
(0.022)
(0.009)
(0.042)
Disaster Exposurei,t-3
Branch * Disaster Exposurei,t-3
Disaster Exposurei,t-3
Branch * Disaster Exposurei,t-3
-0.0172***
-0.0567**
-0.0720***
-0.103**
(0.005)
(0.024)
(0.010)
(0.045)
Disaster Exposurei,t-4
Branch * Disaster Exposurei,t-4
Disaster Exposurei,t-4
Branch * Disaster Exposurei,t-4
-0.0204***
-0.027
-0.0815***
-0.0721
(0.006)
(0.026)
(0.012)
(0.047)
Disaster Exposurei,t-5
Branch * Disaster Exposurei,t-5
Disaster Exposurei,t-5
Branch * Disaster Exposurei,t-5
-0.0218***
-0.00099
-0.142***
-0.0437
(0.007)
(0.027)
(0.013)
(0.050)
Branch * Disaster Exposurei,t-6
Disaster Exposurei,t-6
Branch * Disaster Exposurei,t-6
Disaster Exposurei,t-6
-0.0207***
0.0321
-0.153***
-0.0164
(0.007)
(0.030)
(0.014)
(0.050)
Disaster Exposurei,t-7
Branch * Disaster Exposurei,t-7
Disaster Exposurei,t-7
Branch * Disaster Exposurei,t-7
-0.0360***
0.0113
-0.157***
0.0574
(0.008)
(0.033)
(0.015)
(0.053)
Disaster Exposurei,t-8
Branch * Disaster Exposurei,t-8
Disaster Exposurei,t-8
Branch * Disaster Exposurei,t-8
-0.0239***
0.0232
-0.141***
0.312***
(0.008)
(0.033)
(0.013)
(0.049)
Disaster Exposurei,t-9
Branch * Disaster Exposurei,t-9
Disaster Exposurei,t-9
Branch * Disaster Exposurei,t-9
-0.0111
-0.00607
-0.0979***
0.398***
(0.007)
(0.032)
(0.012)
(0.050)
Disaster Exposurei,t-10
Branch * Disaster Exposurei,t-10
Disaster Exposurei,t-10
Branch * Disaster Exposurei,t-10
-0.0159**
0.0434
-0.0684***
0.291***
(0.007)
(0.031)
(0.011)
(0.048)
Disaster Exposurei,t-11
Branch * Disaster Exposurei,t-11
Disaster Exposurei,t-11
Branch * Disaster Exposurei,t-11
-0.0183**
0.012
-0.0322***
0.348***
(0.007)
(0.030)
(0.010)
(0.044)
Branch * Disaster Exposurei,t-12
Disaster Exposurei,t-12
Branch * Disaster Exposurei,t-12
Disaster Exposurei,t-12
-0.00722
-0.00994
-0.0180*
0.195***
(0.006)
(0.029)
(0.010)
(0.047)
Branchi,j,t-1
Branchi,j,t-1
Log of Bank Assets
-0.0453***
0.00104
Log of Bank Assets
-0.0146***
0.000946
(0.004)
(0.004)
(0.004)
(0.002)
Markets w/o branches
Markets with Branches
Markets w/o branches
Markets with Branches
Coefficient Sum
-0.224
-0.293
Coefficient Sum
-1.088
0.091
F
41.44
5.28
F
305
0.56
P-Value
0.0000
0.022
P-Value
0.0000
0.455
County by Time FE
Bank by County FE
Number of clusters (banks)
Observations
R-squared
*** p<0.01, ** p<0.05, * p

Yes
Yes
5,457
1,609,550
0.257

County by Time FE
Bank by County FE
Number of clusters (banks)
Observations
R-squared

Yes
Yes
2,604
5,726,125
0.397

Table 6: The effect of credit demand shocks on connected markets, Jumbo v. Non-Jumbo Mortgages
This table reports the regression of the change mortgage originations for bank i/county j/month t on the change in
lending in counties hit by natural disasters. The models include County x Month fixed effects (to absorb demand
shocks) and Bank x County fixed effects (to absorb unobserved heterogeneity), with standard errors clustered by
bank. The sample includes all banks with assets above $2 billion. A county is included if the bank originated any
mortgages in the prior year.

Disaster Exposurei,t-1
Disaster Exposurei,t-2
Disaster Exposurei,t-3
Disaster Exposurei,t-4
Disaster Exposurei,t-5
Disaster Exposurei,t-6
Disaster Exposurei,t-7
Disaster Exposurei,t-8
Disaster Exposurei,t-9
Disaster Exposurei,t-10
Disaster Exposurei,t-11
Disaster Exposurei,t-12
Log of Bank Assets
Coefficient Sum
F
P-Value
County by Time FE
Bank by County FE
Number of clusters (banks)
Observations
R-squared
*** p<0.01, ** p<0.05, * p<0.1

Change in Non-Jumbo Mortgage
Originationsi,j,t / Mortgage
originationsi,t

Change in Jumbo Mortgage
Originationsi,j,t / Mortgage originationsi,t

-0.0351***
(0.003)
-0.0323***
(0.004)
-0.0353***
(0.004)
-0.0408***
(0.005)
-0.0481***
(0.005)
-0.0446***
(0.006)
-0.0631***
(0.006)
-0.0406***
(0.006)
-0.0205***
(0.006)
-0.0180***
(0.005)
-0.0162***
(0.005)
-0.00998**
(0.005)
-0.0161***
(0.002)
-0.405
275.4
0.0000

-0.00172*
(0.001)
-0.00533***
(0.001)
-0.00967***
(0.001)
-0.00995***
(0.001)
-0.0117***
(0.002)
-0.0122***
(0.002)
-0.0141***
(0.002)
-0.0107***
(0.002)
-0.0115***
(0.002)
-0.00587***
(0.002)
-0.00407***
(0.001)
-0.000456
(0.001)
-0.00419***
(0.001)
-0.097
159.04
0.000

Yes
Yes
6,273
7,136,703
0.241

Yes
Yes
6,273
7,136,703
0.284

Table 7: The effect of credit demand shocks on connected markets: Jumbo v. Non-Jumbo and Retained v. Sold
This table reports the regression of the change mortgage originations for bank i/county j/month t on the change in lending in counties hit
by natural disasters. The models include County x Month fixed effects (to absorb demand shocks) and Bank x County fixed effects (to
absorb unobserved heterogeneity), with standard errors clustered by bank. The sample includes all banks with assets above $2 billion. A
county is included if the bank originated any mortgages in the prior year.
Change in Non-Jumbo
Change in Non-Jumbo
Sold Mortgage
Retained Mortgage
Originationsi,j,t / Mortgage
Originationsi,j,t /
originationsi,t
Mortgage originationsi,t
Disaster Exposurei,t-1
Disaster Exposurei,t-2
Disaster Exposurei,t-3
Disaster Exposurei,t-4
Disaster Exposurei,t-5
Disaster Exposurei,t-6
Disaster Exposurei,t-7
Disaster Exposurei,t-8
Disaster Exposurei,t-9
Disaster Exposurei,t-10
Disaster Exposurei,t-11
Disaster Exposurei,t-12
Log of Bank Assets
Coefficient Sum
F
P-Value
County by Time FE
Bank by County FE
Number of clusters (banks)
Observations
R-squared
*** p<0.01, ** p<0.05, *
p<0.1

Change in Jumbo
Change in Jumbo
Mortgage Sold
Retained Mortgage
Originationsi,j,t /
Originationsi,j,t / Mortgage
Mortgage originationsi,t
originationsi,t

0.0199***
(0.00375)
0.0167***
(0.00422)
0.0105**
(0.00441)
-0.0105**
(0.00504)
0.0129**
(0.00544)
0.0122**
(0.00594)
-0.00146
(0.00607)
-0.00541
(0.00567)
-0.00299
(0.00577)
0.00463
(0.00534)
0.00458
(0.00519)
0.0193***
(0.00473)
-0.00708***
(0.00136)
0.080
8.82
0.0030

-0.0550***
(0.00322)
-0.0490***
(0.00378)
-0.0459***
(0.00388)
-0.0303***
(0.00432)
-0.0610***
(0.00474)
-0.0568***
(0.00508)
-0.0616***
(0.00533)
-0.0352***
(0.00517)
-0.0176***
(0.00508)
-0.0226***
(0.00490)
-0.0207***
(0.00457)
-0.0293***
(0.00449)
-0.00903***
(0.00145)
-0.485
373.71
0.0000

-0.00456***
(0.00106)
-0.00532***
(0.00133)
-0.00492***
(0.00150)
-0.00483**
(0.00190)
-0.00703***
(0.00203)
-0.00444**
(0.00216)
-0.00747***
(0.00250)
0.00352
(0.00216)
0.00963***
(0.00203)
0.0131***
(0.00192)
0.0117***
(0.00174)
0.0113***
(0.00153)
-0.00360***
(0.000969)
0.011
1.24
0.265

0.00284**
(0.00138)
-9.96e-06
(0.00160)
-0.00474***
(0.00183)
-0.00511**
(0.00224)
-0.00464*
(0.00244)
-0.00772***
(0.00257)
-0.00660**
(0.00296)
-0.0142***
(0.00270)
-0.0211***
(0.00251)
-0.0190***
(0.00240)
-0.0158***
(0.00214)
-0.0117***
(0.00197)
-0.000588
(0.000872)
-0.108
90.68
0.0000

Yes
Yes
6,273
7,136,703
0.24

Yes
Yes
6,273
7,136,703
0.128

Yes
Yes
6,273
7,136,703
0.139

Yes
Yes
6,273
7,136,703
0.154

Table 8: The effect of credit demand shocks on connected markets for Small Banks: Jumbo v. Non-Jumbo and Retained v. Sold
This table reports the regression of the change mortgage originations for bank i/county j/month t on the change in lending in counties
hit by natural disasters for banks operating in fewer than 15 local markets. The models include County x Month fixed effects (to
absorb demand shocks) and Bank x County fixed effects (to absorb unobserved heterogeneity), with standard errors clustered by bank.
The sample includes all banks with assets above $2 billion. A county is included if the bank originated any mortgages in the prior
year.
Change in Non-Jumbo Change in Non-Jumbo
Sold Mortgage
Retained Mortgage
Originationsi,j,t /
Originationsi,j,t /
Mortgage
Mortgage
originationsi,t
originationsi,t
Disaster Exposurei,t-1

Change in Jumbo
Mortgage Sold
Originationsi,j,t /
Mortgage
originationsi,t

Change in Jumbo
Retained Mortgage
Originationsi,j,t /
Mortgage
originationsi,t

Coefficient Sum
F
P-Value

0.0305***
(0.005)
0.0302***
(0.005)
0.0296***
(0.006)
0.00644
(0.007)
0.0401***
(0.007)
0.0445***
(0.008)
0.0307***
(0.008)
0.0267***
(0.008)
0.0260***
(0.008)
0.0323***
(0.007)
0.0240***
(0.007)
0.0362***
(0.006)
-0.0120***
(0.002)
0.357
75.71
0.0001

-0.0575***
(0.004)
-0.0509***
(0.005)
-0.0501***
(0.005)
-0.0277***
(0.006)
-0.0597***
(0.007)
-0.0589***
(0.007)
-0.0618***
(0.008)
-0.0446***
(0.007)
-0.0302***
(0.007)
-0.0379***
(0.007)
-0.0395***
(0.006)
-0.0470***
(0.006)
-0.0282***
(0.002)
-0.566
210.43
0.0000

-0.00229**
(0.001)
-0.00195
(0.001)
-0.00251
(0.002)
0.000634
(0.002)
-0.00354
(0.002)
-0.000952
(0.002)
-0.00754***
(0.003)
-0.000728
(0.003)
0.00391*
(0.002)
0.00639***
(0.002)
0.00363**
(0.001)
0.00462***
(0.001)
-0.0123***
(0.001)
0.000
0
0.9755

0.00437***
(0.002)
0.0014
(0.002)
-0.00214
(0.002)
-0.00403
(0.003)
0.00142
(0.003)
-0.000344
(0.003)
0.0043
(0.004)
-0.0000423
(0.003)
-0.0102***
(0.003)
-0.00873***
(0.003)
-0.00542**
(0.002)
-0.00277
(0.002)
0.00798***
(0.001)
-0.022
2.46
0.1168

County by Time FE
Bank by County FE
Number of clusters (banks)
Observations
R-squared

Yes
Yes
5,313
1,568,928
0.216

Yes
Yes
5,313
1,568,928
0.195

Yes
Yes
5,313
1,568,928
0.296

Yes
Yes
5,313
1,568,928
0.201

Disaster Exposurei,t-2
Disaster Exposurei,t-3
Disaster Exposurei,t-4
Disaster Exposurei,t-5
Disaster Exposurei,t-6
Disaster Exposurei,t-7
Disaster Exposurei,t-8
Disaster Exposurei,t-9
Disaster Exposurei,t-10
Disaster Exposurei,t-11
Disaster Exposurei,t-12
Log of Bank Assets

*** p<0.01, ** p<0.05, * p<0.1

Table 9: The effect of credit demand shocks on connected markets forLarge Banks: Jumbo v. Non-Jumbo and Retained v. Sold
This table reports the regression of the change mortgage originations for bank i/county j/month t on the change in lending in counties
hit by natural disasters for banks operating in more than 15 local markets. The models include County x Month fixed effects (to
absorb demand shocks) and Bank x County fixed effects (to absorb unobserved heterogeneity), with standard errors clustered by bank.
The sample includes all banks with assets above $2 billion. A county is included if the bank originated any mortgages in the prior
year.
Change in Non-Jumbo Change in Non-Jumbo
Sold Mortgage
Retained Mortgage
Originationsi,j,t / Mortgage
Originationsi,j,t /
originationsi,t
Mortgage originationsi,t
Disaster Exposurei,t-1
Disaster Exposurei,t-2
Disaster Exposurei,t-3
Disaster Exposurei,t-4
Disaster Exposurei,t-5
Disaster Exposurei,t-6
Disaster Exposurei,t-7
Disaster Exposurei,t-8
Disaster Exposurei,t-9
Disaster Exposurei,t-10
Disaster Exposurei,t-11
Disaster Exposurei,t-12
Log of Bank Assets
Coefficient Sum
F
P-Value
County by Time FE
Bank by County FE
Number of clusters (banks)
Observations
R-squared
*** p<0.01, ** p<0.05, * p<0.1

Change in Jumbo
Mortgage Sold
Originationsi,j,t /
Mortgage
originationsi,t

Change in Jumbo
Retained Mortgage
Originationsi,j,t /
Mortgage originationsi,t

-0.0212***
(0.007)
-0.00923
(0.007)
-0.0356***
(0.007)
-0.0440***
(0.008)
-0.0530***
(0.008)
-0.0653***
(0.009)
-0.0556***
(0.009)
-0.0520***
(0.008)
-0.0349***
(0.007)
-0.0302***
(0.007)
-0.00865
(0.007)
-0.00246
(0.008)
-0.00578***
(0.002)
-0.412
128.2
0.0000

-0.0451***
(0.007)
-0.0372***
(0.007)
-0.0297***
(0.007)
-0.0238***
(0.007)
-0.0567***
(0.008)
-0.0489***
(0.008)
-0.0613***
(0.008)
-0.0277***
(0.007)
-0.00903
(0.007)
-0.00179
(0.007)
0.0113
(0.007)
0.00533
(0.007)
-0.00498***
(0.001)
-0.325
124
0.0000

-0.00724*
(0.004)
-0.00337
(0.004)
-0.00201
(0.004)
-0.00889*
(0.005)
-0.0142***
(0.005)
-0.00617
(0.006)
-0.0113*
(0.006)
-0.00013
(0.005)
0.0129**
(0.005)
0.0153***
(0.005)
0.0216***
(0.005)
0.0186***
(0.004)
-0.00187*
(0.001)
0.015
0.48
0.4902

-0.00275
(0.004)
-0.0132***
(0.004)
-0.0141***
(0.005)
-0.00956*
(0.005)
-0.0205***
(0.005)
-0.0315***
(0.006)
-0.0210***
(0.007)
-0.0296***
(0.005)
-0.0276***
(0.005)
-0.0239***
(0.005)
-0.0235***
(0.005)
-0.0213***
(0.005)
-0.00198*
(0.001)
-0.239
101.43
0.0000

Yes
Yes
2,599
5,567,775
0.337

Yes
Yes
2,599
5,567,775
0.216

Yes
Yes
2,599
5,567,775
0.142

Yes
Yes
2,599
5,567,775
0.237

Table 10: The effect of credit demand shocks on connected markets for Large Banks: Jumbo v. Non-Jumbo and Retained v. Sold,
with Branch Interactions
This table reports the regression of the change mortgage originations for bank i/county j/month t on the change in lending in counties hit by
natural disasters for banks operating in more than 15 local markets. The models include County x Month fixed effects (to absorb demand
shocks) and Bank x County fixed effects (to absorb unobserved heterogeneity), with standard errors clustered by bank. The sample
includes all banks with assets above $2 billion. A county is included if the bank originated any mortgages in the prior year.
Change in Non-Jumbo Change in Non-Jumbo
Sold Mortgage
Retained Mortgage
Originationsi,j,t /
Originationsi,j,t /
Mortgage originationsi,t Mortgage originationsi,t
Disaster Exposurei,t-1
Disaster Exposurei,t-2
Disaster Exposurei,t-3
Disaster Exposurei,t-4
Disaster Exposurei,t-5
Disaster Exposurei,t-6
Disaster Exposurei,t-7
Disaster Exposurei,t-8
Disaster Exposurei,t-9
Disaster Exposurei,t-10
Disaster Exposurei,t-11
Disaster Exposurei,t-12
Branch * Disaster Exposurei,t-1
Branch * Disaster Exposurei,t-2
Branch * Disaster Exposurei,t-3
Branch * Disaster Exposurei,t-4
Branch * Disaster Exposurei,t-5
Branch * Disaster Exposurei,t-6
Branch * Disaster Exposurei,t-7
Branch * Disaster Exposurei,t-8
Branch * Disaster Exposurei,t-9
Branch * Disaster Exposurei,t-10
Branch * Disaster Exposurei,t-11
Branch * Disaster Exposurei,t-12
Branch
Log of Assets
Coefficient Sum (no branches)
F
P-Value
Coefficient Sum (branches)
F
P-Value
County by Time FE
Bank by County FE
Number of clusters (banks)
Observations
R-squared
*** p<0.01, ** p<0.05, * p<0.1

Change in Jumbo
Mortgage Sold
Originationsi,j,t /
Mortgage
originationsi,t

Change in Jumbo Retained
Mortgage Originationsi,j,t /
Mortgage originationsi,t

-0.0164**
(0.007)
-0.0138*
(0.007)
-0.0314***
(0.007)
-0.0386***
(0.008)
-0.0501***
(0.008)
-0.0635***
(0.009)
-0.0627***
(0.009)
-0.0655***
(0.008)
-0.0544***
(0.007)
-0.0430***
(0.007)
-0.0244***
(0.008)
-0.00877
(0.008)
-0.044
(0.033)
0.0393
(0.027)
-0.0456*
(0.024)
-0.0605**
(0.027)
-0.0353
(0.028)
-0.0216
(0.029)
0.0640**
(0.027)
0.135***
(0.027)
0.200***
(0.025)
0.144***
(0.025)
0.169***
(0.025)
0.0724**
(0.034)
-0.000736
(0.001)
-0.00582***
(0.002)
-0.473
151.4
0.0000

-0.0329***
(0.006)
-0.0296***
(0.006)
-0.0245***
(0.006)
-0.0247***
(0.006)
-0.0569***
(0.007)
-0.0519***
(0.007)
-0.0593***
(0.008)
-0.0406***
(0.007)
-0.0200***
(0.007)
-0.0104
(0.007)
0.00154
(0.007)
0.000665
(0.007)
-0.126***
(0.042)
-0.0867**
(0.035)
-0.0603*
(0.034)
-0.00367
(0.036)
-0.00577
(0.034)
0.0231
(0.035)
-0.0153
(0.037)
0.126***
(0.036)
0.114***
(0.036)
0.0923**
(0.037)
0.0995***
(0.034)
0.046
(0.037)
-0.00113
(0.001)
-0.00496***
(0.001)
-0.349
148.28
0.0000

-0.00783*
(0.004)
-0.00322
(0.004)
-0.00178
(0.005)
-0.00769
(0.005)
-0.0154***
(0.005)
-0.006
(0.006)
-0.0118*
(0.007)
-0.00178
(0.005)
0.0127**
(0.005)
0.0155***
(0.005)
0.0212***
(0.005)
0.0174***
(0.005)
0.00592
(0.006)
-0.0011
(0.007)
-0.00237
(0.007)
-0.0113
(0.008)
0.0107
(0.008)
-0.00142
(0.009)
0.00438
(0.009)
0.0157*
(0.008)
0.00243
(0.008)
-0.00096
(0.007)
0.00476
(0.007)
0.0121
(0.008)
0.00110**
(0.001)
-0.00191*
(0.001)
0.011
0.612
0.6118

-0.00231
(0.004)
-0.0157***
(0.004)
-0.0142***
(0.005)
-0.00984*
(0.006)
-0.0197***
(0.006)
-0.0299***
(0.007)
-0.0222***
(0.007)
-0.0326***
(0.006)
-0.0360***
(0.006)
-0.0285***
(0.005)
-0.0298***
(0.005)
-0.0258***
(0.005)
-0.000789
(0.011)
0.0255**
(0.011)
0.00259
(0.012)
0.00168
(0.014)
-0.00882
(0.014)
-0.0161
(0.015)
0.0096
(0.015)
0.0309**
(0.014)
0.0857***
(0.013)
0.0541***
(0.013)
0.0692***
(0.012)
0.0512***
(0.012)
0.00181***
(0.001)
-0.00207*
(0.001)
-0.267
115.58
0.0000

0.144
2.49
0.1140

-0.145
2.26
0.1330

0.050
3.56
0.059

0.038
0.85
0.3570

Yes
Yes
2,599
5,567,775
0.337

Yes
Yes
2,599
5,567,775
0.216

Yes
Yes
2,599
5,567,775
0.142

Yes
Yes
2,599
5,567,775
0.237