University of Southern California Dissertations and Theses

Understanding the disclosure practices of firms affected by a natural disaster: the case of hurricanes

(USC Thesis Other)

Understanding the disclosure practices of firms affected by a natural disaster: the case of hurricanes

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Understanding the Disclosure Practices of Firms Affected by a Natural Disaster:
The Case of Hurricanes

by

Ventsislav Stamenov

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(BUSINESS ADMINISTRATION)

August 2021

Copyright 2021 Ventsislav Stamenov

ii

Dedication

This dissertation is dedicated to my parents Milko Stamenov and Danka Stamenova,
and my brother Kiril Stamenov. I am thankful for their endless love, encouragement,
and support.
iii

Acknowledgments

I would like to thank my chair Patricia Dechow for the extensive time and energy spent
mentoring me on how to conduct research and answer interesting new questions in
accounting. I am also extremely grateful to Patricia Dechow for shaping me as a
researcher and advising me from the beginning of my research career. I also thank the
remaining member of my dissertation committee Richard Sloan, Jerry Hoberg, and
Forester Wong for their guidance and constructive feedback. In addition, I thank Clive
Lennox, TJ Wong, Lorien Stice-Lawrence, Tom Chang, and workshop participants at the
University of Southern California for helpful feedback on my dissertation. I also thank
the numerous analysts and managers that talked with me about natural disasters. I also
thank my fellow Ph.D. students for their support and feedback throughout the process. I
am grateful to the Marshall School of Business at the University of Southern California
for the financial support.

iv

Table of Contents
Dedication ..................................................................................................................................... ii
Acknowledgments ...................................................................................................................... iii
Table of Contents ........................................................................................................................ iv
List of Tables ................................................................................................................................. v
List of Figures .............................................................................................................................. vi
Abstract........................................................................................................................................ vii
Chapter 1: Introduction ............................................................................................................... 1
Chapter 2: Literature Review ..................................................................................................... 6
Chapter 3: Hypothesis Development ...................................................................................... 13
Chapter 4: Sample and Data ..................................................................................................... 19
Chapter 5: Research Design ...................................................................................................... 27
Chapter 6: Results ...................................................................................................................... 32
Chapter 7: Robustness Tests ..................................................................................................... 37
Chapter 8: Conclusion ............................................................................................................... 41
References ................................................................................................................................... 42
Appendices ................................................................................................................................. 46
Appendix A: Hurricane Characteristics ................................................................................. 46
Appendix B: Randomly Selected Hurricane Dates used for the Placebo Tests ................ 47
Appendix C: Excerpts from Firm Disclosures ....................................................................... 48
Appendix D: Variable Definitions ........................................................................................... 52
v

List of Tables
Table 1: Sample Selection .......................................................................................................... 59
Table 2: Univariate Descriptive Statistics ............................................................................... 63
Table 3: Hurricane Proximity Impact ...................................................................................... 68
Table 4: Market Anticipation During the Hurricane Period for Firms Disclosing
Negative Effect ........................................................................................................................... 70
Table 5: Market Anticipation During the Hurricane Period for Firms Disclosing No
Effect ............................................................................................................................................ 72
Table 6: Disclosure Delay .......................................................................................................... 74
Table 7: Market Reaction During the Disclosure Period for Firms Disclosing No Effect 76
Table 8: Market Reaction Around the Disclosure Period for Firms Disclosing Negative
Effect ............................................................................................................................................ 78
Table 9: Disclosure Salience for Firms Disclosing No Effect ............................................... 80
Table 10: Disclosure Salience for Firms Disclosing Negative Effect ................................... 81

vi

List of Figures

Figure 1: Number of Hurricane Hits from 2003 to 2018 per US County ........................... 54
Figure 2: Aggregate Losses from Hurricanes between 2003 and 2018 per US County ... 55
Figure 3: Historical Hurricane Tracks from 2003 to 2018 ..................................................... 56
Figure 4: Hurricane Timeline for Firms Not Affected by Hurricanes ................................ 57
Figure 5: Hurricane Timeline for Firms Negatively Affected by Hurricanes ................... 58

vii

Abstract

This paper investigates the disclosure practices of firms affected by hurricanes. I
document that when a hurricane hits, there is an increase in investor uncertainty. During
the hurricane period (approximately ten days), there is an increase in abnormal volume,
stock volatility, spread, and illiquidity for firms that later report that they experienced
hurricane damage. I find that firms with little to no impact from the hurricane disclose
this information immediately after the hurricane. In contrast, firms impacted by the
hurricane delay reporting the damage until the next earnings announcement.
Furthermore, firms with “good news” that the hurricane had little damage to operations
disclose this news in the headlines of the earnings release (high salience) while firms that
disclose a negative impact are more likely to bury the news in the body of the earnings
press release (low salience). I also find that hiding the news in the body of the text has
attenuating effect (weaker stock market reaction) on those firms that disclose qualitative
and not quantitative hurricane damage. The results suggest that more timely updates on
hurricane damage to investors reduces stock price volatility.

1

Chapter 1: Introduction

As the effects of global climate change become more prevalent, it will become
increasingly important for investors to obtain timely information on the impact of natural
disasters on the firms’ operations. In this paper, I examine the disclosure practices of firms
affected by hurricanes. I focus on hurricanes because hurricanes are an exogenous shock
that can cause substantial physical and economic damage to firms. As a consequence,
hurricanes adversely affect the firm’s information environment and lead to an increase in
uncertainty for investors. Since firm managers are the only ones that can quickly estimate
the extent of the damage from hurricanes, disclosures practices provide a potentially
important means for them to share their private information with investors. In this paper,
I investigate when and how managers respond to this natural disaster and evaluate
whether certain disclosure practices appear to be more informative than others for
investors.
Hurricanes provide an ideal setting or a “natural experiment” for understanding
disclosure practices of companies for at least three reasons. First, although hurricanes are
not unexpected events in the Florida peninsula, the Gulf States and the East Coast , the
exact timing and path of any hurricane and the degree of damage it may cause cannot be
determined in advance (Belasen and Polachek, 2008). Second, hurricanes are sufficiently
exogenous to the a firm’s characteristics and the extent of damage and losses cannot be
predicted. This is important since this reduces the possibility that variations in a firm’s

2

information environment are due to unobserved heterogeneity or reverse causality
(Dessaint and Matray, 2017). Third, since firms affected by the disaster (treated firms) are
assigned randomly by nature, hurricanes provide a reasonable foundation for drawing
causal inference on the impact of the disaster on the firm’s information environment.
I use textual analysis to identify the firms affected by hurricanes. I search for all
billion-dollar named hurricanes as identified by the Spatial Hazard Events and Losses
Database for the United States (SHELDUS™) between 2003 to 2018. For example, I search
for “Hurricane Katrina” using exact word match in all 10-K, 10-Q and 8-K SEC filings. I
restrict the sample to firms headquartered in the United States and traded on US
Exchanges. I exclude insurance firms and utilities following Michels (2017) because these
firms are either uniquely affected by natural disasters (insurance firms) or they are
regulated industries (utilities). I read the first filing that comes out immediately after the
natural disaster and classify firms into four disclosure types: negatively affected, not
affected, uncertain effect, and positive effect. My final sample of treatment firms consists
of 544 firm-quarters disclosing that they are negatively affected or not affected by a
hurricane. I exclude boildeplate disclosures about uncertain effect and disclosures about
positive effect because there are very few of them (only 18 firm quarters).
The first question I investigate is whether the stock market anticipates the
economic losses related to hurricanes for affected firms. I document that hurricanes are
associated with increased uncertainty for firms that later disclose that they were affected
by the disaster. I also document a similar increase in uncertainty for firms headquartered
in hurricane affected counties. For the ten days from hurricane start to hurricane end, I

3

find statistically significant negative abnormal returns, increases in trading volume, large
increases in bid-ask spreads and increases in Amihud illiquidity for firms negatively
impacted by the hurricane versus a matched sample of firms that have operations in
hurricane affected zones. These results suggest that the market anticipates which firms
will be negatively impacted by the hurricane but there is also increased uncertainty about
the valuation implications.
The second question I investigate is whether firms delay the disclosure of bad
news about the extent of damage from the hurricane. Prior literature is inconclusive as to
whether firms strategically time the disclosure of good versus bad news. For example,
Dye and Sridhar (1995) and Acharya et al. (2011) analytically show that managers tend to
release good news sooner than bad news and that managers cluster their disclosures in
bad times. Similarly, Kothari et al. (2009) argue that a range of incentives, including
career concerns and higher personal wealth stakes, motivate managers to quickly reveal
good news to investors but withhold bad news up to a certain threshold. This finding
contrasts with the conservatism literature (Basu, 1997), which argues that accounting
rules tend to reflect bad news more quickly than good news or that litigation risk can
motivate managers to preempt bad earnings news (Skinner, 1994). The results are
consistent with firms delaying bad news. Firms that are unaffected by the hurricane but
have operations located in the hurricane region, report on average within 20 days of the
start of the hurricane that they were not impacted versus 30 days for firms with a negative
hurricane impact. In addition the firms that report quantifiable dollar damage delay the

4

disclosure of the news longer than firm reporting no effect or firms that report qualitative
damage as it takes them longer to estimate the effect of the damage.
The third question I examine is whether firms benefit from voluntary reporting of
no damage from the hurricane. Firms are not required to disclose information when it is
not material, but I find that there is a group of firms that voluntarily disclose “no damage”
from hurricanes in their press releases. Compared to affected firms, unaffected firms are
usually smaller in size, have less analyst coverage and are less likely to issue management
guidance. Anecdotal evidence from an interview with public relations managers from
one firm suggests that reporting “no damage” from hurricanes is part of its disclosure
practice. Consistent with the predictions of economic theory, I document that these “no
damage” disclosures benefit unaffected firms by decreasing information asymmetry
(Diamond and Verrecchia, 1991). Compared to a Mahalanobis Distance Matched (MDM)
group of firms located in hurricane counties not affected by hurricane, firms reporting
“no damage” from hurricanes have lower abnormal volatility, abnormal spread and
abnormal illiquidity after the disclosure.
The fourth question I explore is the salience of hurricane news in the firms’ press
releases. There are several locations within the press release that firms can disclose
hurricane news: in the headlines or highlights of the earnings release; in the first
paragraph; or in the body of the text. Prior research suggests that higher salience implies
a stronger market reaction (Files et al., 2009; Huang et al., 2018). I find that managers act
strategically: including positive information in the headline but burying more negative
information in the body of the press release. However, unlike Files et al. (2009), I find that

5

the disclosure prominence is not significantly associated with returns after controlling for
the amount of the damage or loss caused by the hurricane. The disclosure prominence is
only statistically negatively associated with stock returns for firms that disclose in their
headlines that they incurred hurricane damage but do not give a dollar amount (i.e., a
negative qualitative disclosure). This result suggests that hiding the negative non-
quantitative news about the hurricane in the body of the text has an attenuating effect,
and is broadly consistent with Files et al. (2009).
Overall, my study contributes to accounting research on disclosure and presents
some new and intriguing results concerning how firms respond to natural disasters. I
document that the market anticipates and does a fairly accurate job at predicting the
economic losses associated with hurricanes. My results also support prior research that
argues that managers disclose good news sooner than bad news. My results extend the
literature on the costs and benefits of voluntary disclosure by providing evidence that
firms not affected by hurricanes still have an incentive to disclose this news. Contrary to
other studies, I also present evidence that although managers may opportunistically
disclose good and bad news information in different parts of the press release, the market
does not appear to ignore or be fooled by hiding information deep in the press release
when the hurricane loss is a dollar amount.

6

Chapter 2: Literature Review

In this chapter, I review the literature related to natural disasters in economics,
finance, and accounting. Although my paper is related to climate change as global
warming is causing an increased intensity of hurricanes (Li and Chakraborty, 2020), I
focus only on these natural disasters used as exogenous shocks and not on the
sustainability or climate change streams of research.
A growing literature in economics, finance, and accounting is estimating the
impact of plausibly exogenous shocks using Difference-in-Difference estimation (DiD),
regression discontinuity (RD), and instrumental variables designs. According to
Atanasov and Black (2016) most of the external shocks documented in the literature are
related to legal rule changes and are often called “natural experiments”. A major issue
related to these natural experiments is whether the shock is exogenous. If firms can
anticipate a large change in the legal environment, then the estimated effect may be much
smaller than the true effect (Hennessy and Strebulaev, 2015). In addition, with more than
two policy states, treatment responses can have incorrect signs and can “overshoot” or
“undershoot” (Hennessy and Strebulaev, 2015).
There is very limited research on how other unexpected events such as natural
disasters affect the economic environment. Natural disasters are so-called acts of God,
while legal changes are intentional. Natural disasters are also different from other
unexpected eventsv such as terrorist attacks which are deliberately caused by humans.

7

Hurricanes, earthquakes, floods, and other natural disasters are primarily acts of Mother
Nature and in some rare cases, a result of human negligence (fires and flooding).
Although in certain regions (such as the Florida peninsula, the Gulf States, and the East
Coast), natural disasters such as hurricanes may be expected events, each hurricane is
exogenous in terms of timing, location, and magnitude of damage. Thus, natural hazards
provide an ideal setting to test how an exogenous change in climate or nature affects the
economic environment.
Research on the impact of hurricanes and natural disasters at the macro level is
already well established. One of the first studies to investigate the economic impact of
natural disasters was Dacy and Kunreuther (1969). They find that GDP generally
increases immediately following a natural disaster. A similar finding is also observed by
Charvériat (2000). He analyzes 35 disaster cases between 1980 and 1996 in Latin America
and the Caribbean to assess their effects on real GDP growth. Charvériat (2000) finds that
real GDP growth tends to decrease in the year of the disaster (by almost 2%) and increase
sharply in the two successive years (by almost 3%). Skidmore and Toya (2002) explain
the reasons for the observed GDP growth after disasters. They find that although disaster
risk reduces the expected rate of return to physical capital, risk also serves to increase the
relative return to human capital. Belasen and Polachek (2008) investigate how hurricanes
affect wages and employment in Florida labor markets. They employ a Generalized
Difference-in-Difference method to examine the effect of hurricanes on the labor market.
This method incorporates many experimental and many control groups to address the
shortcomings of the regular Difference in Difference approach. Belasen and Polachek

8

(2008) compare the changes in employment and earnings for the counties in the state of
Florida affected by 19 hurricanes between 1988 and 2005. The authors find that earnings
of the average worker in a Florida county increase over 4% within the first quarter of
being hit by a Category 4 or 5 hurricane relative to counties not hit and rises about 1.25%
of workers in Florida counties hit by less powerful hurricanes. Interestingly, the effects
of hurricanes on neighboring counties have the opposite effects, causing decreases in
worker earnings of between 3 and 4%. Over time, counties hit by hurricanes experience
a positive net effect on earnings and a negative net effect on employment, but that these
effects dissipate over time.
There are a number of studies that have examined the impact of disasters on banks
and lenders. Some early studies used natural disasters to get exogenous variation in
credit conditions (Morse, 2011; Berg and Schrader, 2012; Cortés and Strahan, 2017). Morse
(2011) finds that after a natural disaster, foreclosures in California increase by 4.5 units
per 1,000 homes. However, the access to payday lenders mitigate 1.0-1.3 of the
foreclosures. Berg and Schrader (2012) use the volcanic eruption in Ecuador to identify
an exogenous increase in loan demand. Their results show that credit demand increases
significantly after volcanic eruptions. Berg and Schrader (2012) show that returning
clients to the Ecuadorean microfinance institution have higher probability of receiving a
loan compared to new clients. Returning clients are also equally likely to be approved for
a loan after the volcanic shock. The results show that bank-borrower relationships are
critical determinants for access to credit in times of unpredictable shocks. Cortés and
Strahan (2017) exploit natural disasters to generate exogenous increases in local credit

9

demand and focus their analysis on connected markets — those where banks lend before
the disaster strikes but are not directly affected. The authors find that banks increase
lending only in markets where they have an informational advantage in the presence of
a branch. At the same time, banks cut lending in connected markets where they have little
or no market power (no branch presence). Thus, banks appear to protect the rents that
they can earn in core markets when they can (Cortés and Strahan, 2017).
There are also some studies that have examined the impact of natural disasters on
publicly traded firms. Barrot and Sauvagnat (2016) investigate whether firm-level
idiosyncratic shocks such as natural disasters propagate in firms’ production networks.
They find that suppliers hit by a natural disaster impose significant output losses on their
customers with an average decrease in sales growth by 2 to 3 percentage points. Barrot
and Sauvagnat (2016) show that these results are economically and significant as the
suppliers represent a small share of firms’ total intermediate inputs. Another study that
examines the impact of hurricanes on firms is the one by Dessaint and Matray (2017).
Their study documents several interesting findings. First, managers of firms located in
the proximity of hurricanes but unaffected by hurricanes respond to the disaster by
increasing the corporate cash holdings by 1.1 percentage points relative to firms farther
away. This effect is also economically significant with an average increase in cash of $15
million and accounting for 10% of the within-firm standard deviation of holdings.
Second, the cash increase is observed only in the first four quarters following the disaster
and then reverts to pre-hurricane levels. Third, cash increases the first and second time a
firm is located in an area neighboring the hurricane counties but not in subsequent

10

occurrences. Dessaint and Matray (2017) explain these results with the availability
heuristics theory. The sudden salience of liquidity risk causes increase in the firm
managers’ perceived liquidity risk (even though the actual risk does not change). These
changes in perception lead managers to increase corporate cash holdings and to express
more concerns about hurricane risk in annual and corporate annual and quarterly filings.
A related study is the one by Ramirez and Altay (2011). They find that firms hoard cash
after a disaster, but this impact is less pronounced in multinational companies.
A study that is related to disclosure and that is most similar to the present stydy
is Michels (2017). Using the setting of subsequent events, Michels (2017) compares the
market response between disclosing and recognizing firms. The study is interesting
because the timing of the natural disaster determines if a firm must disclose or
immediately recognize the financial loss from the disaster. If the natural disaster occurs
prior to the balance sheet date then the firms must immediately recognize the event’s
financial effect. However, if the disaster occurs after the firm’s balance sheet date but
before the firm issues its financial statements (quarterly or annual) the firm must disclose
the financial loss in the footnotes. According to Michels (2017) investors underreact to
disclosed events, consistent with investors incurring higher processing costs when using
disclosed information.
There are several relevant studies that examine the impact of natural disasters on
investors and analysts. Bucciol and Zarri (2013) use the 2010 Health and Retirement Study
(HRS) to provide evidence that there exists correlation between ownership of stocks used
as a proxy for risk aversion and life history negative events out of an individual’s control

11

(having been in a natural disaster and the loss of a child). Bucciol and Zarri (2013) draw
on psychology literature which shows that traumatic events can have wide-ranging and
long-lasting effects on an individual’s dispositions. Their study is interesting as it shows
that risk aversion is correlated only with extreme loss events, such as having been in a
natural disaster and losing a child, but is not correlated with other events, such as having
suffered a severe illness, robbery, or the loss of a job. In a related study, Cameron and
Shah (2015) show in an experiment that individuals who recently suffered a flood or an
earthquake in Indonesia exhibit more risk aversion than individuals living in otherwise
similar villages. Experiencing a natural disaster causes people to perceive that they now
face a greater risk of a future disaster. Similarly, Bernile et al. (2017) show that CEOs who
experience fatal disasters without extremely negative consequences lead firms to behave
more aggressively, whereas CEOs who witness the extreme downside of disasters behave
more conservatively. This non-linear relationship between CEOs past experiences and
firm risk manifest across various corporate policies including leverage, cash holdings,
and acquisition activity (Bernile et al., 2017).
Research on the impact of natural disasters in accounting is still in its infancy.
Bourveau and Law (2016) find that analysts who worked in New Orleans and were
affected by Hurricane Katrina are more pessimistic in their earnings forecasts and
recommendations than analysts outside of Louisiana. The pessimistic analyst forecast are
concentrated in the first 12-18 months after Hurricane Katrina, among stocks with low
analyst competition, and adversely affects firms’ information environment proxied by
forecast dispersion and earnings surprises. Bourveau and Law (2016) explain this analyst

12

behavior with the availability heuristic according to which the ease of recall from memory
leads to systematic errors in judging risk perception (Tversky and Kahneman, 1973).
Building on the availability heuristic, Bourveau and Law (2016) posit that analysts who
recently experienced intense, life-threatening weather events are more likely to associate
tasks involving assessing risk and uncertainty. A related study by Dehaan et al. (2017)
finds that analysts experiencing unpleasant weather conditions are more pessimistic and
less active. Dehaan et al. (2017) show that analysts in locations with unpleasant weather
are slower and less likely to respond to earnings announcements. They draw on the
psychology literature, which characterizes mood as either positive (excited vs. sluggish)
and negative affect (distressed vs. relaxed) (Clark and Watson, 1988). According to
Dehaan et al. (2017) unpleasant weather can trigger negative mood states in analysts such
as fatigue, depression, anxiety, and limited concentration.
Cheng et al. (2019) show that managers of firms that experience natural disasters
are more likely to engage in “big bath” earnings management reporting behavior. They
define big bath as the incidence of negative special items greater than 1% of total assets.
Cheng et al. (2019) confirm that this is an opportunistic behavior as big bath firms have
higher Return on Assets, higher future buy-and-hold returns, and higher executive
compensation in the years following the natural disaster. In a related study, Park (2019)
examines the impact of natural disasters on the likelihood of meeting or beating earnings
benchmarks. Park (2019) finds that firms affected by natural disasters are more likely to
meet-or-beat analyst forecasts through non-GAAP exclusions.

13

Chapter 3: Hypothesis Development

In this chapter, I develop hypotheses on whether and to what extent the stock
market anticipates the types of news reported by firms. My main prediction about
investor uncertainty is based on the model of Kim and Verrecchia (1994). According to
their model, certain informed traders make judgments about a firm’s performance that
are superior to the judgement of other traders. Kim and Verrecchia (1994) show that
expected trading volume is higher and liquidity lower when there is little public
information about a value-relevant event. Applying their model to the hurricane setting
suggests that when a hurricane makes a landfall, certain investors are likely to have a
better idea about which firms have experienced damage (e.g. investors located in the
same geographic region). In addition, there is unlikely to be significant media coverage
of damage effects for specific firms available at the time of the hurricane. As a result, there
will be increased information asymmetry among investors at the time the hurricane
makes a landfall. Furthermore, the unexpected nature of the hurricane news increases
information asymmetry in the market and increases the volatility of prices in the short
run. Hurricane landfall heightens the adverse selection problem and implies that market
makers are at a greater informational disadvantage relative to informed traders (Kim and
Verrecchia, 2001). To protect themselves, the market makers increase the bid-ask spreads
and this makes it more costly for informed traders to trade.

14

It is also possible that as the hurricane strikes, the market price of risk increases,
and so traders reduce the riskiness of their portfolios (Bewley, 2002; Easley and O’Hara,
2010). Using Easley and O’Hara (2010)’s theoretical work, Rehse et al. (2019) show that
Real Estate Investment Trusts (REITs) firms with properties in New York at the time of
Superstorm Sandy have lower trading volume and wider bid-ask spreads than a control
group of firms that do not have properties in New York. Note that the theoretical models
of Kim and Verrecchia (1994) and Easley and O’Hara (2010) provide the same
implications for bid-ask spreads, but the implications for trading volumes differ. Both
models predict increase in the bid-ask spreads (less liquidity) in periods of higher
uncertainty. However, while Kim and Verrecchia (1994) predict higher trading volume
during market uncertainty, Easley and O’Hara (2010) expect decreases in trading volume.
In general, it is natural if the risk is higher, asset prices will fall, and this will generate
more trades (higher volume). However, in their model based on Knightian uncertainty
Easley and O’Hara (2010) allow for both risk and uncertainty to affect market demands.
As Easley and O’Hara (2010) explain “unlike in standard economic models where the
single equilibrium price is an ‘‘average’’ across beliefs, in our uncertainty world there are
a range of market prices”. Although both Easley and O’Hara (2010) and Rehse et al. (2019)
predict lower trading volume in periods of market uncertainty, the Knightian uncertainty
model is applicable mainly to assets in illiquid markets such as the markets for
Collateralized Debt Obligations in the case of Easley and O’Hara (2010) and REITs for
Rehse et al. (2019). Since my study is focused on very liquid equity securities and not on
debt-like securities, I expect hurricanes to be associated with higher bid-ask spreads and

15

higher trading volumes as predicted in Kim and Verrecchia (2001). Based on the above
discussion, I formulate the following hypotheses:
H1a: There is an increase in investor uncertainty during the hurricane period for firms
that have headquarters in hurricane-affected zones.
H1b: There is an increase in investor uncertainty during the hurricane period for firms
that later disclose damage from hurricanes.
For H1a, I focus on firms that have their headquarters located in hurricane regions.
For H1b, I use hindsight to ex-post identify firms that later disclose that they will be
negatively affected by a hurricane.
Prior research documents that managers can strategically time the disclosure of
good and bad news. Both Dye and Sridhar (1995) and Acharya et al. (2011) show that
managers tend to release good news sooner than bad news and that managers are likely
to cluster their disclosures in bad times. А range of incentives, including career concerns
and higher personal wealth stakes, motivate managers to withhold bad news up to a
certain threshold, but quickly reveal good news to investors (Kothari et al., 2009). This
finding is in contrast with the conservatism literature (Basu, 1997), according to which
accounting numbers reflect bad news more quickly than good news. Litigation risk can
also motivate managers to preempt bad earnings news (Skinner, 1994). As a result, firms
often recognize anticipated losses. Based on the discussion above, it is unclear whether
bad news is withheld or disclosed early. I formulate the following hypothesis.
H2a: Firms unaffected by hurricanes that have operations in disaster zones will disclose
the news sooner than firms that are negatively affected by hurricanes.

16

I also expect that firms negatively impacted by hurricanes will delay the disclosure
of “bad news” and the delay will be longer for firms that have material quantifiable
damage consistent with the fact that these firms need more time to evaluate the full extent
of the damage. Based on the above, I formulate the following hypothesis.
H2b: Firms negatively impacted by hurricanes will delay the disclosure of “bad news” and
the delay will be longer for firms that later report quantifiable damage.
Prior literature has shown that usually after a natural disaster like a hurricane,
firms increase their disclosure for risks by expressing more concerns about hurricane risk
in annual (10-Ks) and quarterly reports (10-Qs) (Dessaint and Matray, 2017). This
increased disclosure may last up to six quarters after the occurrence of the natural hazard
event. In addition, firms are encouraged by the SEC (SEC, 2010) to expand their
disclosures if they are affected by natural disasters, especially after Superstorm Sandy
(PwC, 2017). Economic theory suggests that firms have incentives to disclose information
to reduce information asymmetry. This reduction in information asymmetries leads to
increase in stock liquidity and decreased uncertainty (Diamond and Verrecchia, 1991). If
firms disclose more information then there will be less uncertainty about the firm’s
operations (Brown et al., 1987; Fischer and Verrecchia, 2000). Given the preceding
discussion, I formulate the following hypotheses:
H3a: Disclosure by firms located in hurricane zones that they were not affected by the
present hurricane will be associated with decreases in uncertainty for these firms versus a
control group of firms operating in zones unaffected by hurricanes.

17

H3b: Disclosure by firms located in hurricane zones that have negative hurricane damage
will be associated with decreases in investor uncertainty relative to firms located in
hurricane zones that do not update investors about hurricane damage.
Several studies investigate the effect of the press release format on the market’s
reaction. Files et al. (2009) and Gordon et al. (2008) show that managers exercise
considerable discretion over how they announce an accounting restatement and that
disclosure prominence is significantly negatively associated with returns. These studies
document that if managers follow the inverted pyramid axiom from journalism, placing
the most important news (e.g. restatements) at the top, they get penalized with lower
negative returns at the time of the announcement. Managers also tend to present negative
news in a more obscure way. Both Li (2008) and Asay et al. (2018) document that firms’
periodic filings are less readable when the firm’s performance is low, suggesting that
managers have incentives to obfuscate information in order to conceal their
underperformance. Taken together, the arguments suggest that managers will adopt
different disclosure strategies for good versus bad news related to hurricanes.
H4a: Disclosure salience is related to the type of news disclosed by firms. Firms that are
negatively impacted by hurricanes will bury the news in the body of the text, while firms
that are not affected will disclose the “good news” in the headline or first paragraph of the
press release.
In an efficient market, security prices respond quickly to all publicly available
information, so it would not matter if the hurricane news is in the headline or body of the
press release. However, according to Files et al. (2009) limited attention theory

18

(Hirshleifer and Teoh, 2003) predicts that the speed and completeness of price reactions
are both reduced when information is disclosed in a less noticeable format that some
investors may overlook. Based on the discussion above, I formulate the following
hypothesis about the effect of disclosure salience on firm returns:
H4b: Disclosure salience of hurricane damage is negatively associated with returns.

19

Chapter 4: Sample and Data

Appendix A contains a list of all 22 hurricanes from 2003 to 2018. To identify the
hurricanes, I use the named events classified by the Spatial Hazard Events and Losses
Database for the United States (SHELDUS™ version 18.1) as billion-dollar events. The
named events database in SHELDUS™ includes all natural disasters which have been
declared by the President of the United States in the Presidential Disaster Declarations
(PDDs)
1
. I choose hurricanes and not other natural disasters such as tornadoes and fires
because hurricanes are named events that are easily identifiable in the firms’ electronic
filings. In addition, hurricanes are the most damaging of all natural disasters and have
increased in intensity in recent years. For each hurricane, SHELDUS™ provides
information on the start date, the end date and the Federal Information Processing
Standards (FIPS) code of the affected counties. SHELDUS™ has been geocoded to allow
for spatial aggregation and for the property damages to be distributed equally across
affected counties. This results in more conservative estimation of losses compared to the
National Climatic Data Center’s Storm Data database (NCDC). Whenever a president
declares a natural disaster, his actions makes federal funding available to the individuals
in the affected counties. Federal funding is also available to state, tribal, and eligible local
governments and certain nonprofit organizations for emergency work and hazard

1
The disaster declarations by year can be found on the Federal Emergency Management Agency (FEMA)
website: https://www.fema.gov/disasters/year.

20

mitigation measures
2
. Appendix A provides summary statistics for each hurricane,
including start, landfall and dissipation dates, alternate measures of hurricane severity,
the estimated damages, the affected zones, and the number of affected firm headquarters
and operations.
[insert Appendix A here]
Appendix B contains a list of “pseudo” hurricane dates which were used in the
placebo tests. I took the years in which there were no billion-dollar event hurricanes
(2006, 2007, 2013, 2014, and 2015) and created random dates with an average duration of
12 days in the months of August, September, and October, which were the months with
the most significant hurricane activity in the past. I use these “pseudo” hurricane dates
for all my robustness tests.
[insert Appendix B here]
Appendix C contains excerpts from firm disclosures which I read carefully and
categorized in different disclosure types: negatively affected, not affected, positively
affected, and uncertain impact. I focus my analysis on those negatively affected and not
affected disclosures because the positively affected disclosures are very few, and the
uncertain disclosures contain boilerplate language.
[insert Appendix C here]

2
President Donald J. Trump Approves Florida Disaster Declaration:
https://www.whitehouse.gov/briefings-statements/president-donald-j-trump-approves-florida-
disaster-declaration-2/

21

Figure 1 shows the total number of hurricane hits for the US counties for the period
2003-2018. Hurricanes affect a large portion of the US territory. The top five states that
were most often impacted by hurricanes between 2003 and 2018 were Florida, Virginia,
North Carolina, Maryland, and Georgia. The county that experienced the highest number
of hurricanes (12 times) for the period 2003-2018 was Franklin, Florida. I also include the
top ten counties that were hit the most times.
[insert Figure 1 here]
Figure 2 demonstrates the aggregate losses adjusted for 2018 inflation for the US
counties for the period 2003-2018. The top 5 states with the highest cumulative property
damage for this period are Florida, Mississippi, Louisiana, West Virginia and
Connecticut. The county with the highest total loss for the period was Galveston, Texas,
with a combined loss of $13.9 billion. Below Figure 2, I include the top ten counties with
the highest cumulative property losses.
[insert Figure 2 here]
Figure 3 demonstrates the historical hurricane tracks for the Atlantic hurricane for
the period 2003-2018. Hurricanes, also known as cyclones and typhoons in other parts of
the world, form in warm-water oceans near the equator. The Atlantic hurricane season
begins June 1
st
and ends November 30
th
. Hurricanes usually affect the Gulf of Mexico
states, the Florida Peninsula, and the East Coast. The Gulf of Mexico states are Alabama,
Louisiana, Mississippi, Texas, and Florida (excluding the Florida Peninsula), and the East
coast states include North and South Carolina, Virginia, New York, and the New England

22

states. The hurricane formation, landfall location and category are also included in Figure
3.
[insert Figure 3 here]
Panel A of Table 1 summarizes the process for selecting my sample of firms
disclosing that they are affected by hurricanes. I construct my sample by searching firms’
filings on the Securities and Exchange Commission Electronic Data Gathering, Analysis,
and Retrieval (EDGAR) system for the specific name of hurricanes identified by
SHELDUS™ as billion-dollar events. For example, I search for the exact word match
“Hurricane Katrina” using in all 10-K, 10-Q and 8-K SEC filings. I start in 2003 because
from 1995 to 2002 the firms’ disclosures related to hurricanes are scarce and not detailed
enough to perform analyses. The sample includes quarterly data for all U.S. firms with
required data from Compustat, CRSP, and SEC Edgar filings for the period 2003 to 2018.
I require non-missing earnings announcement dates, price greater than $1, and market
capitalization greater than $5 million. I also restrict the sample to firms headquartered in
the United States and traded on US Exchanges. I exclude insurance firms and utilities
following Michels (2017) because these firms are either uniquely affected by natural
disasters (financials) or they are regulated industries (utilities). I read all the disclosures
and keep only disclosures in which firms mention that they were affected by a hurricane
and exclude boilerplate disclosures. My final sample consists of 544 firm-quarter
disclosures (438 unique firms).
[insert Table 1 Panel A here]

23

Panel B of Table 1 summarizes the reconciliation of unique firms to firm quarters.
There are 438 unique firms affected by hurricanes, suggesting that very few of the firms
are impacted more than once by a hurricane. The top three companies that were hit the
most times by hurricanes are Group 1 Automotive Inc, Dillard’s Inc, and Omega Protein
Corp.
[insert Table 1 Panel B here]
Panel C of Table 1 tabulates the four different types of disclosures made by firms
in their SEC filings: negatively affected by hurricane, not affected, uncertain effect, and
positive effect. Negatively affected firms are most likely to disclose while positively
affected firms are least likely to disclose. I exclude uncertain effect firms from my analysis
since it is not clear how hurricanes impact these firms. I also exclude firms that disclose
that they expect a positive effect from the hurricane because of insufficient data (only 18
firm quarter observations for the whole period 2003-2018). Typically, home improvement
retailers, such as Home Depot, Inc. and Lowe’s Companies, Inc, mention that their sales
will be positively affected by the hurricanes.
[insert Table 1 Panel C here]
Panel D of Table 1 shows that the majority of firms disclosing that they are affected
by a hurricane have operations or headquarter locations in a state affected by a hurricane.
To identify firms’ state operations, I hand collect information about the geographical
location of all branches and subsidiaries of the 438 firms that reports that they were
affected by a hurricane from Infogroup’s ReferenceUSA™ U.S. Historical Business
Database. For the firms that do not have data in ReferenceUSA™, I identify firms’ state

24

operations using a measure of geographic dispersion based on state names mentioned in
the annual report filed with the SEC based on Garcia and Norli (2012). Following the
same approach, I count the occurrence of state names in sections “Item 1: Business”, “Item
2: Properties”, “Item 6: Consolidated Financial Data”, and “Item 7: Management’s
Discussion and Analysis” in forms 10-K, 10-K405, 10-KSB, 10-KT, 10KSB, 10KSB40,
10KT405 and the amendments to these forms. I only count states in a form in a given fiscal
year. Firms that mention a state name at least once in their annual report are considered
to have a geographical presence in that state, while firms that do not mention that state
are considered not to have a geographical presence in the state. In addition to all 50 states,
I also include one sovereign territory (Puerto Rico) because hurricane Maria 2017 affected
mainly Puerto Rico. To identify firms’ headquarter location, I use location data obtained
from Loughran and McDonald (2016) to identify firms whose headquarters are affected
by a hurricane. Loughran and McDonald’s dataset captures all of the information in the
header section of the annual and quarterly reports filed on EDGAR and appends some
additional data such as latitude, longitude, and population, which are based on the
business address zip code as reported in the filing.
3
Based on this dataset, I am able to
identify the exact county location for the majority of companies in Compustat. If
Loughran and McDonald’ data is not available, then I use the zip code address from
Compustat. The reason to select Loughran and McDonald (2016) over Compustat is that
Compustat only records the last available location of the headquarters of each firm.

3
I would like to thank Professor Bill McDonald for guiding me on where to find the data and how to merge it with
Compustat. The data is publicly available and includes headquarter location for the period 1996 to 2018
https://sraf.nd.edu/data/augmented-10-x-header-data/

25

[insert Table 1 Panel D here]
Panel E of Table 1 depicts the filing types used by firms to disclose the news related
to the hurricane effect. The majority of the announcements are in the earnings
announcements (8-K Item 2.02 Results of Operations) and the quarterly reports (10-Q
Item 2 Management Discussion and Analysis).
[insert Table 1 Panel E here]
Panel F of Table 1 illustrates that 168 of the firms disclosing negative hurricane
effect have material hurricane loss that impacts revenue, expenses, gross profit, operating
earnings or earnings per share (EPS), while 180 firms disclose only qualitative damage.
[insert Table 1 Panel F here]
Panel G of Table 1 has the tabulation of keywords from firms’ disclosures. The
table shows that the most frequently used terms for firms negatively affected by
hurricanes are sales, expenses, and damages.
[insert Table 1 Panel G here]
Table 2 illustrates the univariate descriptive statistics for the market reaction
during the hurricane period, during the disclosure date, and after the disclosure date. In
Table 2 Panel A for firms disclosing no effect we see slightly negative abnormal returns
and higher abnormal volatility and abnormal spread during the hurricane period.
[insert Table 2 Panel A here]
In table 2 Panel B for firms disclosing negative effect we see significantly negative
returns, abnormal volume, abnormal volatility, and abnormal illiquidity.
[insert Table 2 Panel B here]

26

Table 2 Panel C provides descriptive statistics for the entire sample of firms 14,520.
The main variables of interest are cumulative abnormal return, abnormal volume,
abnormal spread, and abnormal volatility during the hurricane period. The control
variables used in the analysis are size, market to book, leverage, return on assets, loss,
institutional holdings, and earnings surprise.
[insert Table 2 Panel C here]
Table 2 Panel D provides descriptive statistics for the firms disclosing negative
effect and their matched control firms. Table 2 Panel E provides descriptive statistics for
the firms disclosing no effect and their matched control firms.
[insert Table 2 Panel D and Table 2 Panel E here]
Table 2 Panels F, G, and H provide the Pearson and Spearman correlations
between variables for the full sample (Panel F), the negative effect firms and their
respective matches (Panel G), and the firms disclosing no effect (Panel H). The Spearman
correlations are above and Pearson below the diagonal.
[insert Table 2 Panel F, Table 2 Panel G and Table 2 Panel H here]

27

Chapter 5: Research Design
In order to test hypothesis H1a that hurricanes are associated with increased
uncertainty for firms located in hurricane zones, I estimate the following ordinary least
squares regression model:
𝑈𝑛𝑐𝑒𝑟𝑡𝑎𝑖𝑛𝑡𝑦 = 𝛼 0
+ 𝛼 1
𝐻𝑄𝑖𝑛𝐻𝑢𝑟𝑟𝑖𝑐𝑎𝑛𝑒 𝑍𝑜𝑛𝑒 + ∑ 𝛼 𝑖 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑖 +
𝑖 𝜀 (1)
where uncertainty variable is the cumulative abnormal return, abnormal volume,
abnormal volatility, abnormal spread, or abnormal illiquidity. Daily abnormal returns are
estimated using the Fama-French three-factor model with an estimation window from
December 1 to May 31 for each year before the start date of the hurricane, requiring 70
valid daily returns. Cumulative abnormal returns are estimated during the event period,
starting one day before the hurricane start date and ending one day after the hurricane
end date. Abnormal trading volume is the average daily share trading volume in the
event period divided by the average daily share volume in the non-event period.
Abnormal spread is the average bid-ask spread in the event period divided by the
average bid-ask spread in the non-event period. Bid-ask spreads are calculated as in
Corwin and Schultz (2012). Abnormal illiquidity is the average illiquidity in the event
period divided by the average illiquidity in the non-event period. Amihud’s average
illiquidity measure for the event period is calculated with the following formula
𝐴𝑉𝐺𝐼𝐿𝐿𝐼𝑄 𝑖𝑡
= (1 𝑛 𝑡 ⁄ ) × ∑ 𝑅𝐸𝑇 𝑖𝑡
(𝑅𝐸𝑇 𝑖𝑡
× 𝑉𝑂𝐿𝑈𝑀𝐸 𝑖𝑡
) ⁄ where 𝑛 𝑡 is the number of days in
the event period, 𝑅𝐸𝑇 𝑖𝑡
is daily returns, and (𝑅𝐸𝑇 𝑖𝑡
× 𝑉𝑂𝐿𝑈𝑀𝐸 𝑖𝑡
) is the daily dollar
trading volume for stock i on day t.

28

Figures 4 and 5 show a graphical timeline representation of the periods used in the
calculation of the uncertainty variables. Hurricane season in the United States starts on
June 1
st
and ends on November 30
th
. The event period (aka the hurricane period) is the
period from the hurricane start date to hurricane end date and is on average 10 days. The
estimation period is on average 180 days and starts in December 1
st
the year before the
hurricane and ends in May 31
st
in the year of the hurricane. This estimation period is used
to calculate the Fama-French three-factor model returns, average volume, average bid-
ask spread, average volatility, and average Amihud illiquidity. The quiet (non-
disclosure) period is 20 (30) days for firms disclosing no effect (for firms disclosing
damage). The period between the disclosure date and the next earnings announcement
date is on average 44 (53) days for firms disclosing no effect (for firms disclosing
damage).
[insert Figures 4 and Figure 5 here]
HQ in Hurricane Zone is one if a firm’s headquarters is located in a hurricane-
affected county and zero if a firm’s operations are located in hurricane affected by
hurricanes. Controls are size, market-to-book, leverage, return on assets, loss,
institutional holdings, and earnings surprise. Size is measured as the natural log of the
market capitalization. Market-to-Book is the market value of equity divided by the book
value of equity, and both are measured at the beginning of the quarter. Leverage is
current liabilities plus long-term debt as a percentage of total assets, all measured at the
beginning of the quarter. Return on Assets is net income at the beginning of the quarter
as a percentage of total assets. Loss is one if the net income at the beginning of the quarter

29

is negative and zero otherwise.. Institutional holdings is the percent of shares outstanding
held by institutions. Earnings Surprise is the I/B/E/S consensus forecast before the event
date, scaled by the price at the end of the same fiscal quarter. If I/B/E/S data is
unavailable, I use the seasonal random walk earnings surprise using Compustat data.
To test hypothesis H1b that the stock market anticipates the bad news for firms
negatively affected by hurricanes, I estimate the following ordinary least squares
regression model:
𝑈𝑛𝑐𝑒𝑟𝑡𝑎𝑖𝑛𝑡𝑦 = 𝛼 0
+ 𝛼 1
𝐴𝑓𝑓𝑒𝑐𝑡𝑒𝑑 + ∑ 𝛼 𝑖 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑖 +
𝑖 𝜀 (2)
where outcome variables are the same as in (1). Affected is 1 for firms disclosing that they
were negatively affected by hurricanes; and 0 for a matched sample of firms that have
operations in states affected by hurricanes but not making disclosures about hurricanes.
Controls are the same as in (1). To select the sample of matched firms for the firms
disclosing negative effect, I use Mahalanobis Distance Matching (MDM) (Michels, 2017).
I choose the matched firms to have headquarters or operations in hurricane affected
states. The variables used for the MDM are size, market to book, leverage, return on
assets, loss, institutional holdings and earnings surprise. The MDM matching technique
has been shown to outperform propensity score matching in relatively small samples
(Michels, 2017).
To test hypothesis H2a that firms unaffected by hurricanes that have operations in
disaster zones tend to disclose the news sooner than firms that are negatively affected by
hurricanes, I estimate the following Poisson regression model:
𝐷𝑎𝑦𝑠 = 𝛼 0
+ 𝛼 1
𝑈𝑛 𝑎 𝑓𝑓𝑒𝑐𝑡𝑒𝑑 + ∑ 𝛼 𝑖 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑖 +
𝑖 𝜀 (3)

30

where Days is number of days from the hurricane start date to the date of the first
disclosure related to a hurricane. Unaffected is 1 for firms disclosing that they are not
affected by hurricanes and 0 for an MDM matched sample of firms that that disclose that
they are affected by a hurricane.
To test hypothesis H2b that firms that disclose dollar damage by hurricanes take
longer to disclose than firms that disclose negative hurricane effect, I estimate the
following Poisson regression model:
𝐷𝑎𝑦𝑠 = 𝛼 0
+ 𝛼 1
𝑄𝑢𝑎𝑛𝑡𝑖𝑡𝑎𝑡𝑖𝑣𝑒𝐷𝑎𝑚𝑎𝑔𝑒 + ∑ 𝛼 𝑖 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑖 +
𝑖 𝜀 (4)
QuantitativeDamage is 1 for firms disclosing quantitative hurricane loss or damage and 0
for firms disclosing negative hurricane effect. I construct the matched sample using
closest match based Mahalanobis distance metric similar to a normalized Euclidian
distance, but also accounting for correlation between variables (Michels, 2017). I match on
the control variables- size, market to book, leverage, return on assets, loss indicator,
institutional holdings, and earnings surprise. For each disclosing firm, I select with
replacement the firm that is closest based on the Mahalanobis distance of the matching
variables (Michels, 2017). I also used propensity score matching, and the results are
identical. The advantage of Mahalanobis distance over propensity score matching is that
Mahalanobis distance also accounts for correlations between variables and outperforms
propensity score matching in relatively small samples.
Hypothesis H3a predicts that firms disclosing that they are unaffected by a
hurricane have lower abnormal volume, abnormal volatility, and spread than a control

31

group of firms located in states not affected by hurricanes. To test this prediction, I
estimate the following model:
𝑈𝑛 𝑐 𝑒𝑟𝑡𝑎𝑖𝑛𝑡𝑦 = 𝛼 0
+ 𝛼 1
𝑁𝑜𝑡 𝐴𝑓𝑓𝑒𝑐𝑡𝑒𝑑 + ∑ 𝛼 𝑖 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑖 +
𝑖 𝜀 (5)
where uncertainty variables are the same as in (1) but are measured during the disclosure
period. Not Affected is 1 for firms disclosing that they are not affected by hurricanes and
0 for a matched sample group of firms that do not have operations in hurricane-affected
zones. Controls are same as in (1).
To test hypothesis H3b that firms disclosing that they are affected by hurricane
have lower abnormal volume, abnormal volatility and spread than a control group of
firms located in hurricane zones, I estimate the following model:
𝑈𝑛𝑐𝑒 𝑟𝑡𝑎𝑖𝑛𝑡𝑦 = 𝛼 0
+ 𝛼 1
𝐴𝑓𝑓𝑒𝑐𝑡𝑒𝑑 + ∑ 𝛼 𝑖 𝐶𝑜𝑛𝑡𝑟𝑜𝑙 𝑖 +
𝑖 𝜀 (6)
where outcome variable is as in (5) and Affected is 1 for firms disclosing that they are
negatively affected by hurricanes and 0 for a matched sample 1 group of firms that that
have operations in hurricane affected states. Controls are size, market-to-book, leverage,
return on assets, loss, institutional holdings, and earnings surprise.
To test hypothesis H4a I use univariate results to show the difference in the
frequency of firms disclosing hurricane news in the headline and body of the press release
because of the small sample size.
To test hypothesis H4b I follow the research design of Files et al. (2009) and use
univariate results to estimate the difference in means between abnormal returns and
dollar loss hurricane estimates.

32

Chapter 6: Results
Panel A of Table 3 reports both the univariate and multi-variate results of
regression (1) of effects of hurricane proximity of a firm on cumulative abnormal return,
abnormal volume, abnormal volatility, abnormal spread, and abnormal illiquidity during
the hurricane period. Consistent with prediction H1(a) that hurricanes are associated
with increased investor uncertainty, I find that firms whose headquarters are in hurricane
affected counties experience significantly higher abnormal volume, abnormal volatility
and abnormal spread during the hurricane period than firms that have operations in
hurricane affected states. Specifically, firms whose headquarters are located in hurricane
affected zones have 0.5% lower abnormal returns, 8% higher abnormal volume, 24%
higher abnormal volatility and 6% higher abnormal spread than firms who have
operations in hurricane affected zones (for the multi-variate results). These results are
both highly statistically and economically significant.
[insert Table 3 Panel A here]
In Panel B of Table 3, I provide the results of a placebo test that enhance the
credibility of my results and verify that the results are most likely caused by hurricane
events. These results are discussed in Chapter 7.
Tables 4 provides evidence for hypothesis H1b which tests whether there is a
negative market reaction at the time of the hurricane for firms that later report that they
are negatively affected (Panel A) or not affected (Panel B) by the hurricane. The results in
Table 4 Panel A indicate that there is a significant decrease in the cumulative abnormal

33

return and significant increase in the cumulative abnormal volume and abnormal
volatility for firms that later report that they were affected by a hurricane versus a
matched sample of firms which operate in hurricane affected states. Specifically, firms
which later report negative effect have 1.2% lower abnormal returns, 14% higher
abnormal volume, and 3% higher abnormal volatility than firms who have operations in
hurricane affected zones (for the multi-variate results). As expected none of the control
variables are statistically significant in Table 4 as the matching on these variables has
already controlled for much of their variation (Michels, 2017). In Panel B of Table 4, I
provide the results of a placebo test that verify that my results are reliably caused by
hurricane events. These results are discussed in Chapter 7.
[insert Table 4 Panel A here]
I perform similar tests for hypothesis H1b for the firms that later voluntary report
that they were not affected by hurricanes. In Table 5 Panel A the results indicate that the
market correctly anticipates the “good news” as there is no significant change in
abnormal returns, volume, and volatility for firms that later report that they were not
affected by a hurricane compared to a matched group of firms which operate in states not
affected by hurricanes.
[insert Table 5 Panel A here]
Table 6 Panel A provides evidence for Hypothesis H2a. Firms disclosing no effect
report the news sooner than firms disclosing negative effect. On average the firms
disclosing no effect report 10 days sooner than matched firms that disclose a negative
effect. Firms are not required to disclose the news that they were not affected, so this

34

disclosure is voluntary. After speaking to a public relations officer from one firm, I
learned that this firm appears to prefer to disclose no hurricane effect as part of its
disclosure practice.
[insert Table 6 Panel A here]
Table 6 Panel B provides corroborating evidence that firms disclosing a dollar
amount for hurricane damage report on average 8 days later than matched firms
disclosing qualitative damage. The disclosure delay is consistent with the fact that firms
need more time to evaluate the degree of damage and provide an exact estimate of the
dollar damage amount.
[insert Table 6 Panel B here]
Firms are required by mandate of the Securities and Exchange Commission to
disclose information if this information is material. The materiality thresholds, typically
quantitative percentages used by audit firms, are normally 0.5% of total assets, 5% of the
absolute value of net income, and 1.0% of sales (Eilifsen and Messier Jr, 2015). The choice
for the different materiality measures is admittedly arbitrary. In summary, results from
Table 6 Panel A and Panel B indicate that firms delay the disclosure of “bad news.”
However, tests cannot distinguish whether this is strategic or due to actual effects. For
example, the delay could reflect the fact that more time is needed to coordinate with the
auditors or other valuation experts and to confirm the degree of the damage.
I next investigate the results for the consequences of disclosure as hypothesized in
H3a. Panel A of Table 7 reports the market reaction for the firms that are voluntarily
disclosing that they are not affected by a disaster. Results indicate statistically significant

35

decreases in abnormal volume, abnormal volatility, and Amihud illiquidity for the 10
days after the disclosure period for firms disclosing they were not affected versus
Mahalanobis Distance Matched firms located in zones that were not affected by
hurricanes.
[insert Table 7 Panel A here]
I also perform a placebo test for the pseudo hurricane dates in Panel B of Table 7.
The placebo results are explained in detail in Chapter 7.
Panel A of Table 8 reports the results for H3b for firms that report negative effect
from hurricanes versus a matched group of firms located in hurricane affected zones. I
do not see a significant difference between abnormal volume, abnormal volatility and
Amihud illiquidity between negatively affected firms and their matched counterparts. I
also perform a placebo test for the pseudo hurricane dates in Panel B of Table 8. The
placebo results are explained in detail in Chapter 7.
[insert Table 8 Panel A here]
Panel A of Table 9 reports the results for H4a. We observe that firms are 63% more
likely to disclose that they were not affected by hurricanes in the headline or first
paragraph. Panel B of Table 9 demonstrates that firms bundle the news of beating analyst
forecasts (positive surprise) with positive news that they are not affected by a disaster.
Finally, Panel C of Table 9 shows that when firms report positive earnings surprise (beat
analyst forecast) and voluntarily disclose good news that they are not affected, we
observe statistically significant cumulative Fama-French three factor abnormal returns.
[insert Table 9 Panel A, Table 9 Panel B and Table 9 Panel C here]

36

Panel A of Table 10 reports the disclosure salience for firms reporting that they are
negatively affected by hurricanes. These firms are equally likely to disclose the news in
the headline or the body of the press release. Panel B of Table 10 is an evidence that firms
that miss analyst forecast are more likely to include the hurricane news in the body of the
press release while firms that beat analyst forecast are more likely to include the hurricane
news in the headline or the first paragraph. Panel C of Table 10 shows that the market
reacts more to the negative earnings surprise than the disclosure salience. Panel D of
Table 10 illustrates that larger quantitative hurricane losses are more likely to be in the
headlines, but this strategic placement is not associated with differences in abnormal
returns. Panel E illustrates that for firms that disclose qualitative hurricane damage there
is as an attenuation effect. The abnormal returns are statistically negative for firms which
place the news in the headlines and not statistically different from zero for firms that
place the news in the body of the press release. This result is different from Files et al.
(2009) who find that the companies providing medium or low prominence benefit from a
less negative market reaction but for quantifiable restatements.
[insert Table 10 Panels A-E here]
In summary, my results confirm my main hypotheses H1a, H1b, H3a, and H3b
and show that the increased uncertainty during the hurricane period (H1a and H1b) and
the decreased uncertainty during the disclosure period (H3a, and H3b). My results also
confirm the delay of bad news (H2a and H2b) and the strategic disclosure by managers
trying to bundle hurricane news with other good news (beat analyst expectations) and
bad news (miss analyst expectations).

37

Chapter 7: Robustness Tests
In order to verify that all of my results are reliably caused by exogenous hurricane
events and no other confounding factors, I perform placebo tests for all my results. To
achieve this, I randomly assign hurricane dates for years in which there were no
hurricanes (see Appendix A). I keep the average duration of the placebo hurricanes
similar to the average duration of the real hurricanes and I randomly pick dates within
hurricane season, which in the United States starts June 1
st
and ends November 31
st
. The
years that I use in my placebo analysis do not have billion-dollar event hurricanes: 2006,
2007, 2013, 2014, and 2015.
The first placebo test I use is to verify H1a that there is an increase in investor
uncertainty during the hurricane period for firms located in hurricane affected zones.
Panel B of Table 3 reports both univariate and multi-variate results of regression (1) of
effects of hurricane proximity of a firm on cumulative abnormal return, abnormal
volume, abnormal volatility, abnormal spread, and abnormal illiquidity during the
pseudo hurricane period.
The results in Panel B of Table 3 indicate that were no significant decreases in price
or significant increases in abnormal volume, abnormal volatility, and abnormal spread
for the firms headquartered in hurricane zones compared to control firms. The results of
the placebo tests confirm my H1a hypothesis that the increase in uncertainty is likely
caused by the hurricane themselves and not by other random events such as press
releases and earnings announcements.

38

[insert Table 3 Panel B here]
The second placebo test I use is to verify H1b that there is an increase in investor
uncertainty during the hurricane period for firms that later report that they were
negatively affected by hurricanes. Panel B of Table4 reports both univariate and multi-
variate results of regression (2) of a dummy of whether a firm is negatively affected or
not on cumulative abnormal return, abnormal volume, abnormal volatility, and
abnormal volatility during the pseudo hurricane period.
The results in Panel B of Table 4 indicate that were no significant decreases in price
or significant increases in abnormal volume, abnormal volatility, and abnormal volatility
for the firms headquartered in hurricane zones compared to control firms. The results of
the placebo tests confirm my H1b hypothesis that the increase in uncertainty is
attributable to the hurricane events.
[insert Table 4 Panel B here]
The third placebo test I use is to verify if there is an increase in investor uncertainty
during the hurricane period for firms that later report that they were not affected by
hurricanes. Panel B of Table4 reports both univariate and multi-variate results of
regression (2) of a dummy of whether a firm is negatively affected or not on cumulative
abnormal return, abnormal volume, abnormal volatility, and abnormal volatility during
the pseudo hurricane period.
The results in Panel B of Table 5 indicate that were no significant decreases in price
or significant increases in abnormal volume, abnormal volatility, and abnormal volatility
for the firms headquartered in hurricane zones compared to control firms. The results of

39

the placebo tests provide additional support my H1b hypothesis that the increase in
uncertainty during the hurricane period is only for firms that later report that they are
affected by disasters and not for firms that report that they are not affecte.
[insert Table 5 Panel B here]
The fourth placebo test I use is to verify H3a that there is a decrease in investor
uncertainty during the disclosure period for firms that later report that they are not
affected by hurricanes. The Placebo test for the same group of firms in Panel B of Table 7
shows that during the “pseudo hurricane” events there are not statistically significant
differences between the firms reporting no effect and the matched group of firms. The
results confirm my prediction about the benefits of voluntary disclosure for firms
voluntarily reporting “good news”.
[insert Table 7 Panel B here]
The fifth and final placebo test I use is to verify H3b that there is a decrease in
investor uncertainty during the disclosure period for firms that later report that they were
affected by hurricanes. As we saw in the results for real hurricanes, there was no decrease
in uncertainty during the disclosure period. However, in the Placebo test in Panel B of
Table 8, firms that disclose bad news normally have statistically higher abnormal
volume, abnormal volatility and Amihud illiquidity after the “pseudo hurricane” events.
In summary, “bad news” firms that later report negative effect also partially benefit from
the disclosure of hurricane events.
[insert Table 8 Panel B here]

40

Overall, all the placebo tests support my hypotheses H1a, H1b, H3a, and H3b and
show that the increased uncertainty during the hurricane period (H1a and H1b) and the
decreased uncertainty during the disclosure period (H3a, and H3b) is largely caused by
the hurricane events.

41

Chapter 8: Conclusion
This study examines the disclosure practices of firms affected by hurricanes. The
benefit of choosing hurricane events is that hurricanes are random and unexpected events
and it is useful to see whether they are anticipated and correctly priced by the stock
market. I document that the stock market anticipates the good and bad news reported by
firms. I predict and find that firms pre-disclose good news when they are not affected by
hurricanes. I also hypothesize and find that firms benefit from the disclosure of hurricane
news. Firms reporting good news are rewarded with lower abnormal volume and
abnormal volatility. Finally, I document that firms purposefully disclose good news
related to hurricanes typically in the headline of the press release while firms bury the
negative news about hurricane effect in the body of the press release when they miss
analyst estimates.
I make three main contributions to the literature. First, I contribute to the literature
regarding the consequences of uncertainty shocks by providing empirical evidence of
how the stock market reacts to hurricane impact: increased trading volume and increased
bid-ask spread for stocks that later report negative effect. Second, I contribute to the
literature on the timeliness of disclosure of good and bad news by showing that managers
voluntarily pre-disclose good news to reduce the uncertainty associated with natural
disasters. Third, I provide evidence for the corporate practices for press release
prominence that could be of use both to investor relation managers and regulators of the
SEC.

42

References
Acharya, V. V., et al. (2011). "Endogenous information flows and the clustering of
announcements." American Economic Review 101(7): 2955-2979.

Asay, H. S., et al. (2018). "Firm performance, reporting goals, and language choices in
narrative disclosures." Journal of Accounting and Economics 65(2-3): 380-398.

Atanasov, V. A. and B. S. Black (2016). "Shock-based causal inference in corporate
finance and accounting research." Critical Finance Review 5: 207-304.

Barrot, J.-N. and J. Sauvagnat (2016). "Input specificity and the propagation of
idiosyncratic shocks in production networks." The Quarterly Journal of Economics
131(3): 1543-1592.

Basu, S. (1997). "The conservatism principle and the asymmetric timeliness of
earnings1." Journal of Accounting and Economics 24(1): 3-37.

Belasen, A. R. and S. W. Polachek (2008). "How hurricanes affect wages and
employment in local labor markets." American Economic Review 98(2): 49-53.

Belasen, A. R. and S. W. Polachek (2008). "How hurricanes affect wages and
employment in local labor markets." American Economic Review 98(2): 49.

Berg, G. and J. Schrader (2012). "Access to credit, natural disasters, and relationship
lending." Journal of Financial Intermediation 21(4): 549-568.

Bernile, G., et al. (2017). "What doesn't kill you will only make you more risk ‐loving:
Early ‐life disasters and CEO behavior." The Journal of Finance 72(1): 167-206.

Bewley, T. F. (2002). "Knightian decision theory. Part I." Decisions in economics and
finance 25(2): 79-110.

Bourveau, T. and K. Law (2016). "Katrina and analysts." Available at SSRN 2782316.

Brown, L. D., et al. (1987). "Security analyst superiority relative to univariate time-series
models in forecasting quarterly earnings." Journal of Accounting and Economics 9(1):
61-87.

Bucciol, A. and L. Zarri (2013). "Financial risk aversion and personal life history."

43

Cameron, L. and M. Shah (2015). "Risk-taking behavior in the wake of natural
disasters." Journal of Human Resources 50(2): 484-515.

Charvériat, C. (2000). "Natural disasters in Latin America and the Caribbean: An
overview of risk."

Cheng, Y., et al. (2019). "Big Bath Accounting Following Natural Disasters." Available at
SSRN 3305478.

Clark, L. A. and D. Watson (1988). "Mood and the mundane: Relations between daily
life events and self-reported mood." Journal of personality and social psychology 54(2):
296.

Cortés, K. R. and P. E. Strahan (2017). "Tracing out capital flows: How financially
integrated banks respond to natural disasters." Journal of Financial Economics 125(1):
182-199.

Corwin, S. A. and P. Schultz (2012). "A simple way to estimate bid ‐ask spreads from
daily high and low prices." The Journal of Finance 67(2): 719-760.

Dacy, D. C. and H. Kunreuther (1969). "economics of natural disasters; implications for
Federal policy."

Dehaan, E., et al. (2017). "Do weather ‐induced moods affect the processing of earnings
news?" Journal of Accounting Research 55(3): 509-550.

Dessaint, O. and A. Matray (2017). "Do managers overreact to salient risks? Evidence
from hurricane strikes." Journal of Financial Economics 126(1): 97-121.

Diamond, D. W. and R. E. Verrecchia (1991). "Disclosure, liquidity, and the cost of
capital." The Journal of Finance 46(4): 1325-1359.

Dye, R. A. and S. S. Sridhar (1995). "Industry-wide disclosure dynamics." Journal of
Accounting Research 33(1): 157-174.

Easley, D. and M. O’Hara (2010). "Liquidity and valuation in an uncertain world."
Journal of Financial Economics 97(1): 1-11.

Eilifsen, A. and W. F. Messier Jr (2015). "Materiality guidance of the major public
accounting firms." AUDITING: A Journal of Practice & Theory 34(2): 3-26.

Files, R., et al. (2009). "Stealth disclosure of accounting restatements." The Accounting
Review 84(5): 1495-1520.

44

Fischer, P. E. and R. E. Verrecchia (2000). "Reporting bias." The Accounting Review
75(2): 229-245.

Garcia, D. and Ø. Norli (2012). "Geographic dispersion and stock returns." Journal of
Financial Economics 106(3): 547-565.

Gordon, E. A., et al. (2008). "Disclosure credibility and market reaction to restatements."
Unpublished working paper. Temple University.

Hennessy, C. A. and I. A. Strebulaev (2015). Beyond random assignment: Credible
inference of causal effects in dynamic economies, National Bureau of Economic
Research.

Hirshleifer, D. and S. H. Teoh (2003). "Limited attention, information disclosure, and
financial reporting." Journal of Accounting and Economics 36(1-3): 337-386.

Huang, X., et al. (2018). "Headline salience, managerial opportunism, and over-and
underreactions to earnings." The Accounting Review 93(6): 231-255.

Kim, O. and R. E. Verrecchia (1994). "Market liquidity and volume around earnings
announcements." Journal of Accounting and Economics 17(1-2): 41-67.

Kim, O. and R. E. Verrecchia (2001). "The relation among disclosure, returns, and
trading volume information." The Accounting Review 76(4): 633-654.

Kothari, S. P., et al. (2009). "Do Managers Withhold Bad News?" Journal of Accounting
Research 47(1): 241-276.

Li, F. (2008). "Annual report readability, current earnings, and earnings persistence."
Journal of Accounting and Economics 45(2-3): 221-247.

Li, L. and P. Chakraborty (2020). "Slower decay of landfalling hurricanes in a warming
world." Nature 587(7833): 230-234.

Loughran, T. I. M. and B. McDonald (2016). "Textual Analysis in Accounting and
Finance: A Survey." Journal of Accounting Research 54(4): 1187-1230.

Michels, J. (2017). "Disclosure versus recognition: Inferences from subsequent events."
Journal of Accounting Research 55(1): 3-34.

Morse, A. (2011). "Payday lenders: Heroes or villains?" Journal of Financial Economics
102(1): 28-44.

45

Park, J. (2019). "Meeting-or-Beating Earnings Benchmarks: The Effect of Natural
Disasters."

PwC (2017). "Accounting and disclosure implications of Hurricanes Harvey and Irma."
No. US2017-21.

Ramirez, A. and N. Altay (2011). "Risk and the multinational corporation revisited: The
case of natural disasters and corporate cash holdings." Available at SSRN 1772969.

Rehse, D., et al. (2019). "The effects of uncertainty on market liquidity: Evidence from
Hurricane Sandy." Journal of Financial Economics 134(2): 318-332.

SEC (2010). "Commission guidance regarding disclosure related to climate change."
Final Rule 6290.

Skidmore, M. and H. Toya (2002). "Do natural disasters promote long ‐run growth?"
Economic inquiry 40(4): 664-687.

Skinner, D. J. (1994). "Why firms voluntarily disclose bad news." Journal of Accounting
Research 32(1): 38-60.

Tversky, A. and D. Kahneman (1973). "Availability: A heuristic for judging frequency
and probability." Cognitive psychology 5(2): 207-232.

46

Appendices
Appendix A: Hurricane Characteristics

This appendix presents key storm and landfall data for the 22 North Atlantic hurricanes that made landfall
in the U.S. in the period January 2003 to December 2018. The column headings are abbreviations for Name
of the Hurricane (Name), Start Date (Start), Landfall Date (Landfall), End Date (End), Duration in Days
(Days), Category (Cat), Damage (Dam), Number of Headquarters Affected by Hurricane (HQ), Number of
Firm Operations Affected by Hurricane (OP), Affected Zone in the US (Zone). Property damage figures are
in billions of US Dollars restated 2018 values. Category (Cat) is based Saffir-Simpson Hurricane Wind Scale.
Average hurricane duration is 12 days, and the average property damage is $27.6 billion. Source: National
Oceanic and Atmospheric Administration (NOAA) and Wikipedia.

Name Start Landfall End Days Cat Dam HQ OP Zone
Isabel 09/06/2003 09/18/2003 09/20/2003 14 2 3.95 7 21 East Coast
Charley 08/09/2004 08/13/2004 08/15/2004 6 4 11.79 2 7 Florida
Frances 08/24/2004 09/05/2004 09/10/2004 17 2 7.22 4 6 Florida
Ivan 09/02/2004 09/16/2004 09/25/2004 23 3 15.11 13 34 Gulf
Jeanne 09/13/2004 09/26/2004 09/29/2004 16 3 7.3 0 2 Florida
Dennis 07/04/2005 07/10/2005 07/18/2005 14 3 1.89 0 0 Gulf
Katrina 08/23/2005 08/25/2005 08/31/2005 8 3 94.35 45 86 Gulf
Rita 09/18/2005 09/24/2005 09/26/2005 8 3 13.96 15 22 Gulf
Wilma 10/16/2005 10/24/2005 10/27/2005 11 3 14.34 13 25 Florida
Dolly 07/20/2008 07/23/2008 07/27/2008 7 1 1.11 0 0 Gulf
Gustav 08/25/2008 09/01/2008 09/07/2008 13 2 5.97 6 9 Gulf
Ike 09/01/2008 09/13/2008 09/15/2008 14 2 25.72 32 43 Gulf
Irene 08/21/2011 08/27/2011 08/28/2011 7 1 13.58 8 18 East Coast
Lee 09/01/2011 09/01/2011 09/05/2011 4 TS 2.54 2 2 Gulf
Isaac 08/21/2012 08/28/2012 09/03/2012 13 1 2.58 5 13 Gulf
Sandy 10/22/2012 10/29/2012 11/02/2012 11 1 64.74 40 58 East Coast
Matthew 09/28/2016 10/04/2016 10/10/2016 12 1 9.98 6 9 Southeast
Harvey 08/17/2017 08/25/2017 09/02/2017 16 4 123.05 15 26 Gulf
Irma 08/30/2017 09/10/2017 09/13/2017 14 4 49.22 8 18 Florida
Maria 09/16/2017 09/20/2017 10/02/2017 16 4 88.6 2 7 Southeast
Florence 08/31/2018 09/14/2018 09/18/2018 18 4 24.3 27 55 East Coast
Michael 10/07/2018 10/10/2018 10/16/2018 9 5 25.1 13 23 Gulf
Average

12

28 12 22

Sum 263 484

47

Appendix B: Randomly Selected Hurricane Dates used for the Placebo
Tests

This appendix presents the random hurricane dates that were chosen in the years in which there were not
billion-dollar hurricanes. The column headings are abbreviations for Name of the Hurricane (Name), Start
Date (Start), Landfall Date (Landfall), End Date (End), Duration in Days (Days). The random dates for the
hurricanes were chosen during hurricane season in the months of August, September and October. The
average duration days for the hurricanes is 12 and it matches the average duration of the real hurricanes.

Name Start Landfall End Days
Hurricane 2006 10/16/2006 10/24/2006 10/28/2006 12
Hurricane 2007 09/18/2007 09/26/2007 09/30/2007 12
Hurricane 2013 08/21/2013 08/27/2013 08/31/2013 10
Hurricane 2014 09/16/2014 09/20/2014 10/02/2014 16
Hurricane 2015 10/07/2015 10/10/2015 10/16/2015 9

48

Appendix C: Excerpts from Firm Disclosures

This appendix shows excerpts taken from SEC Edgar 10-K/10-Q/8-K filings for firms that disclose that
they were negatively affected or not affected by hurricanes. These types of disclosures were manually read
and classified by the author.
Date Filing
Type
Company
Name
(CIK Code)
Types of
Disclosures
Disclosure Notes
10/28/2008 8-K
Item
2.02
Meritage
Homes Corp
(CIK: 833079)
Negatively
Affected
Deterioration in credit markets and
the effects of Hurricane Ike evident in
declines of 25% in total closings and
29% in net orders, with cancellation
rate of 40%, resulting in lower
revenue, increase in unsold homes
and reduced cash flow from
operations in the quarter.
09/14/2017 8-K
Item
7.01
Spirit Airlines
Inc
(CIK: 1498710)
Negatively
Affected
Hurricane Harvey: Our August
operational performance was
negatively impacted by Hurricane
Harvey. Approximately 10 percent of
our network touches Houston. Our
team members quickly restored our
operations, but many affected areas
suffered major damage and we
anticipate a lingering impact from a
reduction in travel demand to and
from the affected areas. Hurricane
Irma: Our September operations
have, and will continue to be,
impacted by Hurricane Irma.
Through September 13, 2017, we have
canceled 1,255 flights related to the
hurricane. Our team members are
working diligently to restore our
operations.
11/03/201 10-Q
Item 2
International
Paper Co
(CIK: 51434)
Negatively
Affected
Operations were negatively impacted
by approximately $30 million of costs
tied to mill and box plant disruptions
caused by Hurricanes Harvey and
Irma.
11/03/201 10-Q
Item 2
International
Paper Co
(CIK: 51434)
Negatively
Affected
Operations were negatively impacted
by approximately $30 million of costs
tied to mill and box plant disruptions
caused by Hurricanes Harvey and
Irma.

49

Appendix C: Excerpts from Firm Disclosures (cont.)

Date Filing
Type
Company Name
(CIK Code)
Types of
Disclosures
Disclosure Notes
11/07/2005 10-Q
Item 2
Fidelity
Bankshares Inc
(CIK: 1028336)
Negatively
Affected
Hurricane Wilma. On Monday, October
24, 2005, our service area was severely
impacted by hurricane Wilma. We
suffered significant damage to three of
our 48 offices. We currently estimate the
cost of these damages to be
approximately $250,000. At November
4, 2005, 46 offices are fully operational.
For security reasons, we have chosen
not to reopen branch offices until power
has been restored to the branch office.
As power is restored, additional offices
will be opening. Currently, customers
are being directed to the nearest
operational office.
09/12/2005 10-K
Item 7
Cardinal Health
Inc
(CIK: 721371)
Not Affected Recent Developments. In late August
2005, Hurricane Katrina devastated
parts of Louisiana, Mississippi and the
Gulf Coast of the United States. The
Company sustained limited damage to
its facilities in the region and at the time
of the filing of this Form 10-K, all major
facilities in the region were operational.
The damage sustained to the
Company’s facilities will not materially
impact its financial condition or results
of operations. The Company has made
monetary and product donations to the
hurricane relief efforts which will not
have a material impact on its financial
condition or results of operations.
09/15/2008 8-K
Item 8.01
American
Spectrum Realty
Inc (CIK:
1121783)
Not Affected American Spectrum Realty Announces
No Major Damage to Houston
Properties Due to Hurricane Ike.
11/02/2018 10-Q
Item 2
Chemed Corp
(CIK: 19584)
Not Affected While we have significant Roto-Rooter
operations in Florida, North Carolina
and Virginia, we had no significant
casualty losses or business interruptions
as a result of Hurricane Florence or
Hurricane Michael.

50

Appendix C: Excerpts from Firm Disclosures (cont.)

Date Filing
Type
Company Name
(CIK Code)
Types of
Disclosures
Disclosure Notes
09/03/2004 10-Q
Item 2
Fleetwood
Enteprises Inc
(CIK: 314132)
Positively
Affected
We will also continue to pursue other
opportunities, such as sales to
community and park operators.
Additionally, the damage caused by
Hurricane Charley has generated some
short-term demand for manufactured
housing products.
11/29/2011 10-Q
Item 2
Home Depot Inc
(CIK: 354950)
Positively
Affected
The positive comparable store sales for
the third quarter and first nine months
of fiscal 2011 reflect a number of factors.
Our performance in the third quarter of
fiscal 2011 was driven by storm related
sales arising from Hurricane Irene and
early winter weather, as well as strength
in our core departments.
12/09/2003 10-Q
Item 2
Lowe’s Inc (CIK:
60667)
Positively
Affected
Further, purchases to prepare for and to
repair damages caused by Hurricane
Isabel in the mid-Atlantic region
contributed to increased third quarter
sales.
12/05/2012 10-Q
Item 2
Village Super
Market (CIK:
103595)
Positively
Affected
Same store sales increased due to very
high sales in the last week of the quarter
as customers prepared for hurricane
Sandy and higher sales in the two stores
in Maryland, which opened on July 28,
2011 and are now included in same store
sales.
09/29/2005 8-K
Item 7.01
Home Solutions
of America
Positively
Affected
The company played a significant role in
our ability to respond quickly to the
need to cleanup the devastation brought
on by the 2004 Florida hurricanes. Due
to the significant increase in hurricane-
related work in the Gulf Coast region
and the acquisition of the assets from
FERS, the Company raised its business
forecast for the second time since mid-
August. It now expects full-year
revenue of $70 million to $75.0 million,
an increase from previous expectations
for revenue of $57.0 to $61.0 million.

51

Appendix C: Excerpts from Firm Disclosures (cont.)

Date Filing
Type
Company Name
(CIK Code)
Types of
Disclosures
Disclosure Notes
11/29/2012 10-K
Item 7
Mueller Water
Pruducts Inc (CIK:
1350593)
Uncertain
Impact
Hurricane Sandy inflicted significant
damage in the northeastern United
States, particularly upon making
landfall on October 29, 2012. The
operations at our facilities in North
Kingstown, Rhode Island and
Columbia, Greencastle and
Waynesboro, Pennsylvania were
temporarily interrupted by the
effects of the hurricane. It is too early
to quantify any potential impact,
whether favorable or unfavorable,
Hurricane Sandy may have on our
results.
09/07/2005 10-Q
Item 2
ADC
Telecommunications
Inc
(CIK: 61478)
Uncertain
Impact
Although we are not certain about
the effect that Hurricane Katrina may
have on sales of our products and
services, it is possible that we will
experience slower sales in the near
term while affected customers work
to stabilize their networks and
normalize operations. Moving
forward, there may also be a
temporary upturn in our sales as our
customers work to replace damaged
or destroyed network elements in the
areas impacted by the Hurricane.
11/06/2012 10-Q
Item 2
Citigroup Inc (CIK:
831001)
Uncertain
Impact
Subsequent Events. Hurricane Sandy
On October 29 and 30, 2012, the
metropolitan New York City region
and New Jersey suffered severe
damage from Hurricane Sandy. Citi
continues to assess the impact on
Citi's facilities and customers in the
affected areas and what impact, if
any, the storm could have on its
results of operations for the fourth
quarter of 2012.

52

Appendix D: Variable Definitions

Variable Description Source
Estimation Window The estimation window for every hurricane year
starts on December 1 one year before the
hurricane start date and ends on May 31 in the
same year as the hurricane start date. Hurricanes
typically occur in the months of August,
September and October.
Wikipedia
Event Window The event window starts one day before the
hurricane start date and ends depending on the
event window used (see Figure 4 and Figure 5). If
the event window is hurricane days then the start
date is t-1 before hurricane start date and the end
date is t+1 after the hurricane end date. If the
event window is quiet, then the start date is t-1
before the hurricane start date and the end date t-
1 before the disclosure date. If the event window
is disclosure days, then the start date is t-1 before
the disclosure date, and the end date is t+1 after
the disclosure date.
Sheldus
CAR Daily abnormal return estimated using the Fama-
French three-factor model with an estimation
window from December 1 to May 31 for each year
before the start date of the hurricane, requiring 70
valid daily returns. Cumulative abnormal returns
are estimated during the event period which
starts one day before the hurricane start date and
ends depending on the event window used.
CRSP
Abnormal Volume Abnormal trading volume is the average daily
share trading volume in the event period divided
by the average daily share volume in the non-
event period.
CRSP
Abnormal Volatility Abnormal volatility is the average squared
abnormal market adjusted returns in the event
period divided by the variance of abnormal
returns in the non-event period.
CRSP
Abnormal Spread Abnormal spread is the average bid-ask spread in
the event period divided by the average bid-ask
spread in the non-event period. Bid-ask spreads
are calculated as in Corwin and Shultz (2012).
CRSP

53

Appendix D: Variable Definitions (cont.)
Variable Description Source
Abnormal Illiquidity Abnormal illiquidity is the average illiquidity
in the event period divided by the average
illiquidity in the non-event period. Amihud’s
average illiquidity measure for the event
period is calculated with the following formula
(1 𝑛 𝑡 ⁄ ) ∑ 𝑅𝐸𝑇 𝑖𝑡
(𝑅𝐸𝑇 𝑖𝑡
× 𝑉 𝑂𝐿𝑈𝑀𝐸 𝑖𝑡
) ⁄
where 𝑛 𝑡 is the number of days in the event
period, 𝑅𝐸𝑇 𝑖𝑡
is daily returns, and
(𝑅𝐸𝑇 𝑖𝑡
× 𝑉𝑂𝐿𝑈𝑀𝐸 𝑖𝑡
) is the daily dollar trading
volume for stock i on day t.
CRSP
HQ in Hurricane Zone 1 if the firm headquarters is located in
hurricane affected county and 0 for all other
firms that have operations in states affected by
hurricanes.
Sheldus
Size the size of company i in year t, measured as the
natural log of market capitalization.
Compustat
Market-to-Book the market value of equity divided by the book
value of equity, and both are measured at the
beginning of the quarter.
Compustat
Leverage current liabilities plus long-term debt as a
percentage of total assets, all measured at the
beginning of the quarter.
Compustat
Return on Assets net income at the beginning of the quarter as a
percentage of total assets.
Compustat
Loss 1 if the net income at the beginning of the
quarter is negative and 0 otherwise.
Compustat
Institutional holdings the percent of shares outstanding held by
institutions.

Earnings Surprise Forecast error scaled by price as of the end of
the same fiscal quarter. The forecast error is
normally based on the most recent I/B/E/S
consensus forecast prior to the event date. If
I/B/E/S data is unavailable, I use the seasonal
random walk earnings surprise using
Compustat data.
I/B/E/S

54

Figure 1: Number of Hurricane Hits from 2003 to 2018 per US County
This figure shows the total number of hurricane hits for the US counties for the period 2003-2018. There
was a total of 22 “billion-dollar” hurricanes that were included in the Spatial Hazard Events and Losses
Database for the United States (SHELDUS). The names of the top ten most impacted counties are listed
below.

County State Number of Hits
Franklin Florida 12
Broward Florida 11
Wakulla Florida 11
Taylor Florida 10
Bay Florida 9
Gulf Florida 9
Dixie Florida 8
Jefferson Florida 8
Miami-Dade Florida 8
Baldwin Alabama 7

55

Figure 2: Aggregate Losses from Hurricanes between 2003 and 2018 per
US County
This figure shows the aggregate losses (property damages and not crop damages) adjusted for 2018
inflation for the US counties for the period 2003-2018. There was a total of 22 “billion-dollar” hurricanes
that were included in the Spatial Hazard Events and Losses Database for the United States (SHELDUS).
The names of the top ten most impacted counties are listed below. The losses from all the counties in Puerto
Rico are combined into one county.

County State Property Damage
Galveston Texas $13,852,937,356
Harris Texas $13,848,931,621
Puerto Rico Puerto Rico $12,921,019,335
Monmouth New Jersey $11,553,617,960
Fort Bend Texas $10,873,704,635
Ocean New Jersey $10,846,879,831
Montgomery Texas $9,516,852,660
Jefferson Louisiana $5,834,285,663
Orleans Louisiana $5,802,457,722
Plaquemines Louisiana $5,768,306,186

56

Figure 3: Historical Hurricane Tracks from 2003 to 2018
This figure shows the historical hurricane tracks for the period 2003-2018. There was a total of 22 “billion-
dollar” hurricanes that were included in the Spatial Hazard Events and Losses Database for the United
States (SHELDUS). The 22 hurricanes were Isabel (2003); Charley, Frances, Ivan and Jeanne (2004); Dennis,
Katrina, Rita and Wilma (2005); Dolly, Gustav and Ike (2008); Irene and Lee (2011); Isaac and Sandy (2012);
Matthew (2016); Harvey, Irma and Maria (2017); Florence and Michael (2018). Source: National Oceanic
and Atmospheric Administration (NOAA).

57

Figure 4: Hurricane Timeline for Firms Not Affected by Hurricanes
This figure shows the historical timeline for firms not affected by hurricanes. Hurricane season in the
United States starts on June 1 and ends on November 30. The estimation period is on average 180 days and
starts in December 1 the year before the hurricane and ends in May 31 in the year of the hurricane. This
estimation period is used to calculate the Fama-French three-factor model returns, average volume, average
bid-ask spread, average volatility and average Amihud illiquidity. The Hurricane Period is on average
around 10 days. The quiet (non-disclosure) period for not affected firms is on average 20 days. The period
between the disclosure date and the next earnings announcement date for not affected firms is on average
44 days.

58

Figure 5: Hurricane Timeline for Firms Negatively Affected by
Hurricanes
This figure shows the historical timeline for firms negatively affected by hurricanes. Hurricane season in
the United States starts on June 1 and ends on November 30. The estimation period is on average 180 days
and starts in December 1 the year before the hurricane and ends in May 31 in the year of the hurricane. This
estimation period is used to calculate the Fama-French three-factor model returns, average volume, average
bid-ask spread, average volatility and average Amihud illiquidity. The Hurricane Period is on average
around 10 days. The quiet (non-disclosure) period for affected firms is on average 30 days. The period
between the disclosure date and the next earnings announcement date for affected firms is on average 53
days.

59

Table 1: Sample Selection
Panel A: Compustat and Hurricane Hit Disclosure Sample

Compustat Sample Disclosure Sample
Sample Selection Firm-
quarters
Unique
firms
Firm-
quarters
Unique
firms
Quarterly observations for US companies
01/01/2003 to 12/31/2018
718,331 23,291 12,945 1,852
Non-missing earnings announcement
date, price greater than $1 and market
capitalization greater than $5 mln.

390,787 14,582 8,294 1,136
Firms traded on US Exchanges

250,273 8,280 6,923 972
Excluding Insurance Carriers (SIC code
6300-6399) and Utilities (4900-4999)
235,768 7,932 4,586 839
Excluding duplicate firm disclosures
related to the same hurricane event

1,244 839
Non-missing information for returns,
volume, spread, and control variables

544 438
Final Sample 235,768 7,932 544 438

60

Table 1: Sample Selection (cont.)

Panel B: Reconciliation of Unique firms to Firm Quarters

Company Name # Firms # Quarters # Firm-Quarters
Group 1 Automotive, Inc. 1 5 5
Dillard’s, Inc. 1 4 4
Omega Protein Corporation 1 4 4
Callon Petroleum Company 1 3 3
Cracker Barrel Old Country Store, Inc. 1 3 3
Diamond Offshore Drilling, Inc. 1 3 3
Encompass Health Corporation 1 3 3
International Paper Company 1 3 3
Noble Energy, Inc. 1 3 3
Pride International, Inc. 1 3 3
Rent-A-Center, Inc. 1 3 3
Valaris plc 1 3 3
Walgreens Boots Alliance, Inc. 1 3 3
Number of Firms Hit Twice 76 2 152
Number of Firms Hit Once 349 1 349
Total Number of Firms and Firm-Quarters 438

544

Panel C: Type of News Disclosed by Firms by Firm-Quarter.

Disclosing Hurricane News Firm-quarters Percent Cumulative
Disclose Negatively Affected 348 63.97% 63.97%
Disclose Not Affected 125 22.98% 86.95%
Disclose Uncertain Effect 53 9.74% 96.69%
Disclose Positive Effect 18 3.31% 100.00%
Total 544 100%

Panel D: Disclosure by Firms Hit by the Hurricane (Affected) vs Other Firms Not in Hurricane
Regions.

Headquarters or State Operations
Located in Hurricane Affected State

YES NO Total
Disclosing Hurricane News 484 60 544
Not Disclosing Hurricane News 18,485 199,386 217,871
Total 18,969 199,446 235,768

61

Table 1: Sample Selection (cont.)

Panel E: Disclosure by SEC Filing Type

Filing Type
Disclosure Type
Negatively
Affected
Not
Affected
Uncertain
Effect
Positively
Affected Total
8-K Item 1.01 Entry into a
Material Definitive Agreement
1 0 0 0 1
8-K Item 2.02 Results of
Operations
165 24 13 9 211
8-K Item 7.01 Regulation FD
Disclosure
46 22 6 1 75
8-K Item 8.01 Other Events 29 28 6 0 63
8-K Item 9.01 Financial
Statements and Exhibits
14 3 0 0 17
10-Q Item 2 Management
Discussion and Analysis
87 43 27 7 164
10-K Item 7 Management
Discussion and Analysis
6 5 1 1 13
Total 348 125 53 18 544

Panel F: Disclose A Quantitative Estimate of the Hurricane Loss vs Qualitative Disclosure

Quantitative or
Qualitative Damage
Disclosure Type
Negatively
Affected
Not
Affected
Uncertain
Effect
Positively
Affected Total
Quantitative Damage 168 4 0 0 172
Qualitative Damage 180 121 53 18 372
Total 348 125 53 18 544

62

Table 1: Sample Selection (cont.)

Panel G: Most Frequently Used Disclosure Terms (Consolidated Synonym Words)

Disclosure Terms Firm-quarters Percent
Disclose Not Affected Firms
• no material impact 125 18.55%
Negatively Affected Firms
• sales 125 18.55%
• expenses 82 12.17%
• damages 69 10.24%
• net income 47 6.97%
• shutdowns 44 6.53%
• disruptions 34 5.04%
• operating income 26 3.86%
• production 23 3.41%
• loan loss provisions 11 1.63%
• unable to file 10-Q 4 0.59%
• power blackout 3 0.45%
• inventory 2 0.30%
• class action lawsuit for oil spill 1 0.15%
• delinquencies 1 0.15%
• employee assistance 1 0.15%
• loan payment extension 1 0.15%
• loans past due 1 0.15%
• waived fees 1 0.15%
• write-offs 1 0.15%
Uncertain Effect
• CEO and CFO unable to attend conference 1 0.15%
• disruptions 1 0.15%
• uncertain impact 52 7.72%
Positively Affected Firms
• sales 18 2.67%
Total 674 100.00%

63

Table 2: Univariate Descriptive Statistics
Table 2 reports firm-level average values for key variables for different periods after the hurricane start
date. Panel A shows average values for key variables for stocks disclosing that they are not affected by
hurricanes. Panel B presents average values for key variables for stocks disclosing that they are affected by
hurricanes. ∗∗∗, ∗∗, and ∗ denote values significantly different from zero at the 1%, 5%, and 10% levels,
respectively.
Panel A: Average Values for Firms Not Affected Firms (N=125)

From t-1 to
t+1 around
Hurricane
Start and
End Date
t-1 before
Hurricane
Start to t-3
before
Disclosure
Date
t-1 before
to t+1
after the
Disclosur
e Date
t+3 to t+15
after the
Disclosure
Date
t+3 after
the
Disclosure
Date to t-2
before the
Next EAD
Days 10.512*** 20.112*** 3.888*** 9.136*** 44.455***
CAR -0.014* -0.013 0.002 -0.016** -0.041***
Abnormal Volume 0.013 0.198 0.49*** 0.151* 0.164**
Abnormal Volatility 1.102*** 1.214*** 2.061*** 1.194*** 1.422***
Abnormal Spread -0.111*** -0.089** -0.062 -0.091 -0.064
Abnormal Illiquidity 0.019 0.089 0.021 0.172 0.510***

Panel B: Average Values for Firms Negatively Affected Firms (N=324)

t-1 before
Hurricane
Start to
t+1 after
Hurricane
End
t-1 before
Hurricane
Start to t-3
before
Disclosur
e Date
t-1 before
to t+1 after
the
Disclosure
Date
t+3 to t+15
after the
Disclosure
Date
t+3 after
the
Disclosure
Date to t-2
before the
Next EAD
Days 11.006*** 30.391*** 3.931*** 9.187*** 52.646***
CAR -0.014*** -0.029*** -0.010** -0.006* -0.031***
Abnormal Volume 0.098*** 0.150*** 0.672*** 0.250*** 0.169***
Abnormal Volatility 1.274*** 1.373*** 5.065*** 1.590*** 1.506***
Abnormal Spread -0.001 0.033 0.215*** -0.025 -0.028
Abnormal Illiquidity 0.115*** 0.218*** 0.384*** 0.343*** 1.689

64

Table 2: Univariate Descriptive Statistics (cont.)

Panel C: Summary Statistics for the full sample of affected firms (N=14,520)

Mean P25 P50 P75
CAR-Hurr 0.009 -0.031 0.003 0.042
Abnormal Volume-Hurr 0.133 -0.393 -0.137 0.239
Abnormal Spread-Hurr 0.061 -0.309 -0.086 0.228
Abnormal Volatility-Hurr 1.259 0.371 0.676 1.284
Size 6.344 4.910 6.272 7.656
Market to Book 1.889 1.074 1.411 2.132
Leverage 0.163 0.004 0.113 0.261
Return on Assets 0.008 0.002 0.012 0.029
Loss 0.216 0.000 0.000 0.000
Institutional Holdings 0.589 0.326 0.642 0.851
Earnings Surprise 0.022 -0.001 0.001 0.004
Note: The full sample in Panel C consists of affected firms have headquarters or operations in a state
affected by hurricanes. The location of firm headquarters is obtained from SEC Edgar and/or Compustat.
The location of firm operations is from either ReferenseUSA or Garcia and Norli (2012)’s measure of
geographic dispersion.

Panel D: Summary Statistics for the 348 Firms Disclosing Negative Effect and their Matched
Control Firms (N=514)

Mean P25 P50 P75
CAR-Hurr -0.009 -0.040 -0.007 0.024
Abnormal Volume-Hurr 0.074 -0.328 -0.086 0.200
Abnormal Volatility-Hurr 1.193 0.428 0.756 1.399
Size 7.448 6.274 7.320 8.613
Market to Book 1.590 1.114 1.382 1.810
Leverage 0.216 0.071 0.196 0.322
Return on Assets 0.019 0.005 0.016 0.029
Loss 0.074 0.000 0.000 0.000
Institutional Holdings 0.727 0.578 0.777 0.876
Earnings Surprise -0.003 0.000 0.001 0.002
Note: The sample in Paneld D consists of firms disclosing negative effect and their matched sample of
control firms. The control firms are obtained from the full sample affected firms.

65

Table 2: Univariate Descriptive Statistics (cont.)

Panel E: Summary Statistics for the 96 Firms Disclosing No Effect and their Matched Control
Firms (N=192)

Mean P25 P50 P75
CAR-Hurr 0.004 -0.033 -0.001 0.033
Abnormal Volume-Hurr -0.008 -0.353 -0.151 0.181
Abnormal Volatility-Hurr 1.178 0.352 0.659 1.326
Size 6.748 5.559 6.922 8.061
Market to Book 1.563 1.070 1.226 1.703
Leverage 0.187 0.032 0.140 0.284
Return on Assets 0.015 0.003 0.009 0.025
Loss 0.146 0.000 0.000 0.000
Institutional Holdings 0.633 0.421 0.715 0.835
Earnings Surprise 0.000 -0.001 0.001 0.003
Note: The sample in Paneld E consists of firms disclosing no effect and their matched sample of control
firms. The control firms are obtained from the sample of Compustat firms that are not affected and not
disclosing.

67
Table 2: Univariate Descriptive Statistics (cont.)
Panel F: Correlation Table for the Full Sample (N=14,520). Spearman above diagonal /Pearson below

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
CAR-Hurr 1 0.11*** 0.03*** 0.07*** 0.09*** 0.02*** 0.02*** 0.04*** -0.03*** 0.08*** 0.01*
Abnormal Volume-Hurr 0.14*** 1 0.2*** 0.37*** 0.18*** 0.09*** 0.04*** 0.1*** -0.07*** 0.1*** 0.01
Abnormal Spread-Hurr 0.03*** 0.07*** 1 0.37*** 0.04*** 0.02** 0.01 0.02** 0 0.07*** -0.03***
Abnormal Volatility-Hurr 0.01* 0.07*** 0.1*** 1 0.03*** -0.04*** 0 0 -0.01 0.02** -0.04***
Size 0.01 -0.06*** 0.02** -0.01 1 0.37*** 0.23*** 0.36*** -0.26*** 0.64*** -0.03***
Market to Book 0.01* 0.02*** -0.01 -0.01 0.21*** 1 -0.16*** 0.37*** 0.01 0.23*** 0.03***
Leverage 0.02** 0 0 -0.01 0.16*** -0.13*** 1 -0.02** -0.06*** 0.18*** -0.01*
Return on Assets 0.04*** -0.05*** 0.01 0 0.27*** -0.08*** -0.01 1 -0.7*** 0.26*** 0.1***
Loss 0.01 0.03*** 0 0 -0.26*** 0.1*** 0.02** -0.62*** 1 -0.14*** -0.1***
Institutional Holdings 0.02* -0.05*** 0.05*** 0 0.61*** 0.09*** 0.17*** 0.21*** -0.14*** 1 -0.04***
Earnings Surprise 0.04*** 0.01 -0.02** -0.01 -0.04*** 0.01 -0.01 0 0 -0.04*** 1

Panel G: Correlation Table for the 348 Firms Disclosing Negative Effect and their Matched Control Firms (N=514)

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
CAR-Hurr 1 -0.06 0.01 -0.04 0.07* -0.03 0 0.02 0.02 0 -0.05
Abnormal Volume-Hurr -0.04 1 0.24*** 0.41*** 0.13*** 0.09** -0.02 0.05 0.03 -0.02 -0.08*
Abnormal Spread-Hurr 0.06 0.2*** 1 0.29*** -0.04 0.03 0 0.06 0.08* -0.02 -0.03
Abnormal Volatility-Hurr -0.22*** 0.35*** 0.22*** 1 0.04 0.03 -0.03 0.02 0.05 0.03 0.01
Size 0.05 -0.01 -0.05 0 1 0.35*** 0.1** 0.21*** -0.1** 0.24*** -0.09**
Market to Book -0.05 0.01 0.01 -0.03 0.22*** 1 -0.01 0.57*** -0.06 0.14*** -0.06
Leverage -0.02 -0.06 -0.02 -0.06 0.04 -0.11** 1 -0.07* 0.05 0.1** 0
Return on Assets -0.08* 0.01 0.02 0.04 0.18*** 0.41*** -0.11*** 1 -0.45*** 0.14*** -0.01
Loss 0.01 0.02 0.09** 0.04 -0.1** -0.07 0.06 -0.45*** 1 -0.01 -0.02
Institutional Holdings 0.01 -0.15*** -0.02 0.01 0.31*** 0.14*** 0.08* 0.12*** -0.01 1 -0.01
Earnings Surprise -0.02 0.07 0 0.06 0.03 0.05 0.06 0.08* -0.04 0.01 1

68
Table 2: Univariate Descriptive Statistics (cont.)

Panel H: Correlation Table for the 96 Firms Disclosing No Effect and their Matched Control Firms (N=192)
Spearman above diagonal /Pearson below

(1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11)
CAR-Hurr 1 0.02 0.06 0.03 0.01 0.13* 0.06 0.07 0.01 0.07 0.12*
Abnormal Volume-Hurr 0.19*** 1 0.14** 0.37*** 0.15** 0.18** -0.07 0.19*** -0.08 0.1 0.12*
Abnormal Spread-Hurr 0.01 0.03 1 0.35*** 0.11 0.12* 0.19*** 0.14** 0 0.15** -0.05
Abnormal Volatility-Hurr 0.1 0.29*** 0.27*** 1 0.04 0.12* 0.11 0.08 0.05 0.07 -0.01
Size 0.04 0.01 0.11 -0.01 1 0.42*** 0.2*** 0.41*** -0.25*** 0.63*** 0.03
Market to Book 0.03 0.09 0.09 0.08 0.36*** 1 -0.01 0.65*** -0.19*** 0.34*** 0.14*
Leverage 0.06 -0.06 0.14** 0 0.15** -0.08 1 -0.01 0.1 0.23*** -0.05
Return on Assets 0.02 0.06 0.16** 0.01 0.39*** 0.63*** -0.04 1 -0.6*** 0.29*** 0.18***
Loss 0.03 0.01 -0.04 0.09 -0.27*** -0.17** 0.18*** -0.49*** 1 -0.08 -0.21***
Institutional Holdings 0.11 0.01 0.11 0 0.66*** 0.26*** 0.19*** 0.26*** -0.1 1 -0.03
Earnings Surprise 0.09 0.1 -0.01 0 0.07 0.09 -0.01 0.19*** -0.31*** 0.08 1***

68

Table 3: Hurricane Proximity Impact
Panel A: Hurricane Proximity Impact for the 22 Hurricanes for the period 2003-2018
Panel A of Table 3 reports the results of the regression of effects of hurricane proximity of a firm on
cumulative abnormal return, abnormal volume, abnormal volatility, abnormal spread, and abnormal
illiquidity for the 22 billion-dollar hurricanes for the period 2003-2018. HQ in Hurricane Zone is 1 if a firm
headquarters is located in hurricane-affected county and 0 for all other firms that have operations in states
affected by hurricanes. All continuous control variables are winsorized at 1 and 99%. Size is the natural
logarithm of the market capitalization at the beginning of the quarter. Market-to-Book is the market value of
equity divided by the book value of equity, and both are measured at the beginning of the quarter. Leverage
is long-term debt scaled by its total assets over the past quarter. Return on Assets is net income over the past
quarter as a percentage of total assets. Loss is 1 if net income at the beginning of the quarter is negative and
0 otherwise. Institutional holdings is the percentage of institutional holdings. Earnings surprise is the
analyst forecast error scaled by price over the last quarter.
Dependent Variable: CAR Abnormal
Volume
Abnormal
Volatility
Abnormal
Spread

(1) (2) (3) (4)
Univariate Results
HQ in Hurricane Zone -0.005*** 0.079*** 0.238*** 0.059***
(-3.30) (2.91) (3.11) (4.55)
Multivariate Results
HQ in Hurricane Zone -0.005*** 0.063** 0.229** 0.051***

(-2.60) (1.98) (2.38) (3.54)
Size -0.001*** -0.015* -0.046 -0.005

(-3.24) (-1.83) (-1.21) (-1.45)
Market to Book 0.000 0.049*** -0.037** -0.006*

(0.71) (4.69) (-2.38) (-1.69)
Leverage 0.008 0.119 -0.177 -0.044

(1.64) (1.50) (-1.62) (-1.52)
Return on Assets -0.082*** -1.092 -0.301 0.072

(-2.66) (-1.45) (-0.62) (0.60)
Loss -0.002 -0.054 -0.069 0.015

(-0.69) (-1.03) (-0.48) (0.89)
Institutional Holdings 0.006** -0.198*** 0.155 0.129***

(2.41) (-4.08) (1.36) (6.55)
Earnings Surprise 0.009** 0.088 -0.100 -0.030*

(2.03) (1.18) (-1.24) (-1.95)
Constant 0.015*** 0.236*** 1.531*** 0.019

(4.90) (4.48) (4.05) (0.94)

Observations 14,520 14,520 14,520 14,520
R-squared 0.005 0.007 0.001 0.005

69

Table 3: Hurricane Proximity Impact (cont.)
Panel B: Placebo Test for Hurricane Proximity Impact
Panel B of Table 3 reports the results of a regression of effects of hurricane proximity of a firm on cumulative
abnormal return, abnormal volume, abnormal volatility, abnormal spread, and abnormal illiquidity for the
random “pseudo” hurricane dates created in Appendix A for years 2006, 2007, 2013, 2014, and 2015. HQ
in Hurricane Zone is 1 if a firm headquarters is located in hurricane-affected county and 0 for all other firms
that have operations in states affected by hurricanes.

Dependent Variable: CAR Abnormal
Volume
Abnormal
Volatility
Abnormal
Spread

(1) (2) (3) (4)
Univariate Results

HQ in Hurricane Zone 0.000 -0.071 -0.093 0.013
(-0.09) (-1.44) (-1.37) (0.86)
Multivariate Results
HQ in Hurricane Zone -0.001 -0.029 -0.146 0.019

(-0.31) (-0.62) (-1.17) (0.98)
Size -0.001 -0.024*** -0.002 -0.002

(-1.40) (-2.81) (-0.22) (-1.40)
Market to Book 0.003*** 0.048*** 0.057*** 0.000

(5.18) (4.55) (2.80) (-0.17)
Leverage 0.0128** 0.019 -0.011 -0.027

(2.48) (0.27) (-0.07) (-1.56)
Return on Assets 0.019 -0.578 -1.311 0.006

(0.98) (-1.22) (-1.45) (0.08)
Loss 0.002 -0.047 -0.091 -0.043***

(1.41) (-1.42) (-1.44) (-5.23)
Institutional Holdings -0.008*** -0.293*** -0.115* -0.0211*

(-3.40) (-6.16) (-1.65) (-1.90)
Earnings Surprise -0.004*** 0.015 0.021 0.007*

(-2.59) (0.60) (0.44) (1.66)
Constant -0.006* 0.334*** 1.166*** -0.013

(-2.52) (4.82) (14.07) (-0.99)

Observations 46,823 45,213 45,210 45,213
R-squared 0.005 0.007 0.002 0.002

70

Table 4: Market Anticipation During the Hurricane Period for Firms
Disclosing Negative Effect
Panel A: Market Anticipation During the Hurricane Period for 348 Firms Disclosing Negative
Effect versus Matched Control Firms
Table 4 reports the results of regression of cumulative abnormal return, abnormal volume abnormal
volatility, abnormal spread, and abnormal illiquidity on whether a firm has disclosed that it has been
affected by hurricane and controls such as size, market to book, leverage, return on assets, loss indicator,
institutional holdings, and earnings surprise. All continuous variables are winsorized at 1 and 99%. Affected
is 1 if a firm later discloses that it was affected by a hurricane and 0 for all other firms that have operations
in states affected by hurricanes. Leverage is long-term debt scaled by its total assets over the past quarter.
Return on Assets is net income over the past quarter as a percentage of total assets. Loss is 1 if net income at
the beginning of the quarter is negative and 0 otherwise. Institutional holdings is the percentage of
institutional holdings. Earnings surprise is the analyst forecast error scaled by price over the last quarter.
Dependent Variable: CAR Abnormal
Volume
Abnormal
Volatility

(1) (2) (3)
Univariate Results
Affected -0.020*** 0.177** 0.108**
(-3.61) (2.28) (2.44)
Multi-variate Results
Affected -0.012** 0.139** 0.033**

(-2.445) (2.13) (2.14)
Size -0.004 -0.009 -0.080

(-1.74) (-0.29) (-1.63)
Market to Book 0.006 -0.016 -0.156

(0.91) (-0.29) (-1.34)
Leverage -0.007 -0.511 -1.405

(-0.27) (-1.74) (-1.65)
Return on Assets -0.432 -0.172 -6.216

(-1.14) (-0.07) (-0.94)
Loss 0.008 -0.096 -0.125

(0.43) (-0.72) (-0.38)
Institutional Holdings 0.020 0.013 0.270

(1.52) (0.06) (0.82)
Earnings Surprise 0.004 0.557* 0.508*
(0.13) (1.80) (1.75)
Constant 0.025 0.167 2.363***

(1.31) (0.76) (3.92)

Observations 514 514 514
R-squared 0.055 0.018 0.043

71

Table 4: Market Anticipation During the Hurricane Period for Firms
Disclosing Negative Effect (cont.)
Panel B: Placebo Test for Market Anticipation During the Hurricane Period for 348 Firms
Disclosing Negative Effect versus Matched Control Firms
Table 4 reports the results of regression of cumulative abnormal return, abnormal volume abnormal
volatility, abnormal spread, and abnormal illiquidity on whether a firm has disclosed that it has been
affected by hurricane and controls such as size, market to book, leverage, return on assets, loss indicator,
institutional holdings, and earnings surprise. All continuous variables are winsorized at 1 and 99%.
Dependent Variable: CAR Abnormal
Volume
Abnormal
Volatility

(1) (2) (3)
Univariate Results
Affected -0.004 0.026 -0.059
(-1.57) (0.60) (-0.79)
Multi-variate Results
Affected -0.005 -0.006 -0.020

(-1.27) (-0.17) (-0.28)
Size 0.001* -0.009 0.008

(1.70) (-0.83) (0.55)
Market to Book 0.002 0.057** 0.152

(0.86) (2.09) (1.58)
Leverage -0.001 -0.137 -0.310

(-0.17) (-1.23) (-1.27)
Return on Assets 0.132 1.620 -1.255

(1.13) (1.47) (-0.70)
Loss 0.005 0.011 -0.122

(0.91) (0.17) (-1.15)
Institutional Holdings -0.010* -0.384*** 0.131

(-1.66) (-4.01) (0.66)
Earnings Surprise -0.010 0.017 -0.006
(-1.01) (0.24) (-0.06)
Constant -0.014* 0.293*** 0.835***

(-1.96) (2.82) (4.83)

Observations 2,340 2,340 2,340
R-squared 0.010 0.017 0.006

72

Table 5: Market Anticipation During the Hurricane Period for Firms
Disclosing No Effect
Panel A: Market Anticipation During the Hurricane Period for 96 Firms Disclosing No Effect
Table 5 reports the results of regression of cumulative abnormal return, abnormal volume and abnormal
volatility on whether a firm has disclosed that it is not affected by a hurricane. All continuous variables are
winsorized at 1 and 99%. Not Affected is 1 if a firm later discloses that it was not affected by a hurricane and
0 for all other firms that have operations in states not affected by hurricanes. Leverage is long-term debt
scaled by its total assets over the past quarter. Return on Assets is net income over the past quarter as a
percentage of total assets. Loss is 1 if net income at the beginning of the quarter is negative and 0 otherwise.
Institutional holdings is the percentage of institutional holdings. Earnings surprise is the analyst forecast
error scaled by price over the last quarter. Note: Excludes firms disclosing during the Hurricane Period.
Dependent Variable:
CAR
Abnormal
Volume
Abnormal
Volatility
(1) (2) (3)
Univariate Results
Not Affected -0.004 -0.140 -0.297
(-0.35) (-0.10) (-1.06)
Multi-variate Results
Not Affected -0.013 -0.006 -0.007

(-1.15) (-0.05) (-0.03)
Size -0.006 -0.057 0.077

(-1.42) (-1.17) (0.99)
Market to Book -0.017 0.334*** 0.520**

(-1.49) (2.85) (2.09)
Leverage -0.056 -0.046 0.625

(-1.431) (-0.120) (0.64)
Return on Assets -0.413 1.565 -6.828***

(-0.990) (0.87) (-3.261)
Loss -0.019 -0.208 -0.626**

(-0.773) (-0.856) (-2.405)
Institutional Holdings 0.079*** -0.417* -1.351**

(3.33) (-1.746) (-2.252)
Earnings Surprise 0.004 0.557* 0.408
(0.13) (1.80) (1.45)
Constant 0.038 0.158 0.799*

(1.40) (0.50) (1.70)

Observations 192 192 192
R-squared 0.125 0.111 0.119

73

Table 5: Market Anticipation During the Hurricane Period for Firms
Disclosing No Effect (cont.)
Panel B: Placebo Test for Market Anticipation During the Hurricane Period for 96 Firms
Disclosing No Effect
Table 5 reports the results of regression of cumulative abnormal return, abnormal volume and abnormal
volatility on whether a firm has disclosed that it is not affected by a hurricane and controls such as size,
market to book, leverage, return on assets, Property Plant & Equipment, loss indicator, institutional
holdings and earnings surprise. All continuous variables are winsorized at 1 and 99%. Excludes firms
disclosing during the Hurricane Period.
Dependent Variable:
CAR
Abnormal
Volume
Abnormal
Volatility

(1) (2) (3)
Univariate Results
Not Affected 0.002 0.048 -0.049
(0.40) (0.58) (-0.40)
Multi-variate Results
Not Affected 0.003 0.033 -0.059

(0.60) (0.36) (-0.50)
Size -0.003 -0.015 0.107

(-1.10) (-0.40) (1.49)
Market to Book 0.002 0.043 -0.031

(0.43) (0.70) (-0.30)
Leverage 0.025 -0.129 -0.742

(1.17) (-0.50) (-1.33)
Return on Assets 0.491** 1.802 0.659

(1.98) (0.59) (0.20)
Loss 0.006 0.048 0.444

(0.58) (0.31) (0.87)
Institutional Holdings 0.010 -0.428** -0.389*

(1.00) (-2.30) (-1.86)
Earnings Surprise -0.022 0.056 1.319
(-1.05) (0.25) (1.03)
Constant -0.010 0.426 0.785***

(-0.76) (1.61) (2.70)

Observations 930 930 930
R-squared 0.028 0.013 0.036

74

Table 6: Disclosure Delay
Panel A: Disclosure Delay for Firms Disclosing No Effect vs Firms Disclosing Negative Effect
Table 6 reports the results of Poisson regression of disclosure delay on whether a firm has disclosed that it
is not affected versus a Mahalanobis distance matched firms that have disclosed that they are affected by
hurricane. All continuous variables are winsorized at 1 and 99%. Not Affected is 1 if a firm later discloses
that it was not affected by a hurricane and 0 for firms that later disclose that they were affected by a
hurricane. Leverage is long-term debt scaled by its total assets over the past quarter. Return on Assets is net
income over the past quarter as a percentage of total assets. Loss is 1 if net income at the beginning of the
quarter is negative and 0 otherwise. Institutional holdings is the percentage of institutional holdings. Earnings
surprise is the analyst forecast error scaled by price over the last quarter. Note: Excludes firms disclosing
during the Hurricane Period.
Dependent Variable: Disclosure Delay

(1)
Univariate Results
Not Affected -0.362***
(-4.34)
Multi-variate Results
Not Affected -0.357***

(-4.34)
Size 0.046

(1.58)
Market to Book 0.017

(0.24)
Leverage 0.111

(0.40)
Return on Assets -1.840

(-0.89)
Loss -0.057

(-0.41)
Institutional Holdings -0.181

(-1.03)
Earnings Surprise 0.152

(0.57)
Constant 3.622***

(19.20)

Observations 206

75

Table 6: Disclosure Delay (cont.)

Panel B: Disclosure Delay for Firms Disclosing Quantitative Dollar Damage vs Firms Disclosing
Qualitative Damage
Table 6 reports the results of Poisson regression of disclosure delay on whether a firm has disclosed
quantitative damage versus a Mahalanobis distance matched firm that have disclosed qualitative damage.
Multivariate results include controls for size, market to book, leverage, return on assets, loss indicator,
institutional holdings and earnings surprise. All continuous variables are winsorized at 1 and 99%. Firms
disclosing quantitative damage report on average 8 days later than firms disclosing no quantitative
damage.

Dependent Variable: Disclosure Delay

(1)
Univariate Results
Quantitative Dollar Damage 0.194***

(2.78)
Multi-variate Results

Quantitative Dollar Damage 0.196***

(2.83)
Size -0.020

(-0.80)
Market to Book -0.013

(-0.33)
Leverage 0.082

(0.34)
Return on Assets 0.529

(0.33)
Loss -0.015

(-0.09)
Institutional Holdings 0.061

(0.38)
Earnings Surprise -0.133

(-0.90)
Constant 3.812***

(18.80)

Observations 228

76

Table 7: Market Reaction During the Disclosure Period for Firms
Disclosing No Effect
Panel A: Market Reaction During 10 days after the Disclosure Date for Firms Disclosing No
Effect for the 22 hurricanes from 2003 to 2018.
Panel A of Table 7 reports the results of regression of abnormal volume, abnormal volatility, and abnormal
illiquidity on whether a firm has disclosed that it is affected by a hurricane and controls such as size, market
to book, loss, institutional holdings and earnings surprise. All continuous variables are winsorized at 1 and
99%.

Dependent Variable: Abnormal
Volume
Abnormal
Volatility
Abnormal
Spread
Amihud
illiquidity

(1) (2) (3) (4)
Univariate Results
Not Affected -0.09** -0.27** -0.08** -0.25**

(-2.18) (-2.35) (-1.97) (-1.99)
Multi-variate Results

Not Affected -0.10** -0.23** -0.06** -0.24**

(-2.26) (-2.53) (-1.97) (-1.96)
Size 0.0844* -0.230 -0.025 -0.189*

(1.69) (-0.65) (-0.51) (-1.70)
Market to Book -0.015 -0.458 0.037 -0.112

(-0.22) (-1.59) (0.69) (-1.19)
Leverage -0.258* -0.708 -0.009 0.117

(-1.89) (-0.97) (-0.05) (0.33)
Return on Assets -0.532 2.995 -0.092 1.020

(-1.32) (1.01) (-0.36) (1.32)
Loss 0.016 0.320 0.019 -0.541

(0.07) (0.50) (0.12) (-1.56)

Constant 0.052 2.454* 0.139 1.186**

(0.13) (1.66) (0.59) (2.41)

Observations 206 206 206 206
R-squared 0.034 0.062 0.013 0.066

77

Table 7: Market Reaction During the Disclosure Period for Firms
Disclosing No Effect (cont.)

Panel B: Placebo Test for Market Reaction During 10 days after the Disclosure Date for Firms
Disclosing No Effect
Panel B reports the results of regression of cumulative abnormal return, abnormal volume abnormal
volatility, abnormal spread, and abnormal illiquidity on whether a firm has disclosed that it was not
affected by a hurricane and controls such as size, market to book, leverage, return on assets, Property Plant
& Equipment, Loss indicator and institutional holdings. All continuous variables are winsorized at 1 and
99%.

Dependent Variable: Abnormal
Volume
Abnormal
Volatility
Abnormal
Spread
Amihud
illiquidity

(1) (2) (3) (4)
Univariate Results
Not Affected 0.142 -0.168 0.113 0.061

(1.62) (-0.877) (1.38) (0.95)
Multi-variate Results
Not Affected 0.116 -0.114 0.131 0.083

(1.29) (-0.646) (1.45) (1.31)
Size 0.041 0.160 0.041*** -0.031

(1.19) (1.19) (2.90) (-1.62)
Market to Book -0.043 -0.248** -0.054** 0.037

(-0.85) (-2.02) (-2.40) (0.98)
Leverage -0.452 -1.622 -0.277*** 0.055

(-1.65) (-1.57) (-2.87) (0.30)
Return on Assets 0.121 0.813 0.046 0.264

(0.78) (0.97) (0.84) (1.65)
Loss -0.233 -0.171 -0.020 -0.011

(-1.32) (-0.68) (-0.25) (-0.07)
Constant 0.214 2.512 -0.040 -0.489

(0.87) (0.94) (-0.58) (-1.63)

Observations 968 968 968 968
R-squared 0.009 0.043 0.032 0.025

78

Table 8: Market Reaction Around the Disclosure Period for Firms
Disclosing Negative Effect
Panel A: Market Reaction During the 10 days after the Disclosure Date for Firms Disclosing
Negative Effect for the 22 Hurricanes for the period 2003-2018
Panel A of Table 8 reports the results of regression of cumulative abnormal return, abnormal volume
abnormal volatility, abnormal spread, and abnormal illiquidity on whether a firm has disclosed that it was
not affected by a hurricane and controls such as size, market to book, leverage, return on assets , Loss
indicator and institutional holdings. All continuous variables are winsorized at 1 and 99%.

Dependent Variable: Abnormal
Volume
Abnormal
Volatility
Amihud
illiquidity

(1) (2) (4)
Univariate Results
Affected -0.120 0.202 0.176

(-0.99) (0.90) (1.09)
Multi-variate Results
Affected -0.119 0.201 0.176

(-1.00) (0.90) (1.09)
Size -0.029 0.080 -0.035

(-0.59) (1.15) (-1.40)
Market to Book -0.030 -0.088 0.026

(-0.79) (-0.71) (0.34)
Leverage 0.165 0.610 0.482**

(0.78) (1.49) (2.37)
Return on Assets -0.334 1.054 0.711***

(-1.36) (1.29) (3.06)
Loss 1.488 1.260* 0.069

(1.10) (1.68) (0.14)
Constant 0.878 0.176 -0.275

(1.55) (0.52) (-1.48)

Observations 556 556 556
R-squared 0.028 0.022 0.053

79

Table 8: Market Reaction Around the Disclosure Period for Firms
Disclosing Negative Effect (cont.)

Panel B: Placebo Test for Market Reaction During 10 days after the Disclosure Date for Firms
Disclosing Negative Effect
Panel B of Table 8 reports the results of regression of cumulative abnormal return, abnormal volume
abnormal volatility, abnormal spread, and abnormal illiquidity on whether a firm has disclosed that it was
not affected by a hurricane and controls such as size, market to book, leverage, return on assets, Loss
indicator and institutional holdings. All continuous variables are winsorized at 1 and 99%.

Dependent Variable: Abnormal Volume Abnormal
Volatility
Amihud
illiquidity

(1) (2) (4)
Univariate Results
Affected 0.027*** 0.121*** 0.101***

(2.48) (7.09) (3.17)
Multi-variate Results
Affected 0.034** 0.127*** 0.113***

(2.85) (7.31) (3.45)
Size -0.010 0.003 0.000

(-0.38) (0.41) (0.01)
Market to Book -0.049 0.004 0.041

(-0.92) (0.27) (1.58)
Leverage -0.510* -0.118* -0.188

(-1.95) (-1.79) (-1.59)
Return on Assets -0.132 -0.056* 0.137*

(-1.01) (-1.71) (1.85)
Loss -0.001 0.111** -0.009

(-0.00) (2.40) (-0.07)
Constant -0.224 -0.047 -0.056

(-1.64) (-1.33) (-0.63)

Observations 2,626 2,626 2,626
R-squared 0.018 0.022 0.008

80

Table 9: Disclosure Salience for Firms Disclosing No Effect
Panel A of Table 9 provides descriptive statistics for the disclosure prominence of the hurricane news in
the press release (8-K report). Not affected firms are more likely to disclose the “good news” in the headline
or the first paragraph. Panel B shows the disclosure salience for firms by earnings surprise. Firms are more
likely to bundle positive earnings surprise and the news that they are not affected in the Headline or the
first paragraph of the 8-K. Panel C reports the market reaction for the disclosure of no hurricane damage
and whether the firm beats the market expectation. The market reaction measured as the abnormal Fama-
French three factor adjusted return is positive and significantly different from zero for firms that beat
market expectations and disclose in the new of no hurricane damage in the headline of the press release.

Panel A: Disclosure of “No Hurricane Damage” in the Firm Press Release

Disclosure Salience Number Percent of Firms
Headline or First Paragraph 40 63%
Body of the Press Release 24 38%
Total 64 100%

Panel B: Disclosure of “No Hurricane Damage” and Whether A Firm Beats or Misses Analysts’
Earnings Expectations

Disclosure Salience
Beats
Expectations
Percent Misses
Expectations
Percent
Headline or First Paragraph 28 64% 10 50%
Body of the Press Release 16 36% 10 50%
Total 44 100% 20 100%

Panel C: Disclosure of “No Hurricane Damage” and Stock Market Returns
Disclosure Salience Beats Expectations Misses Expectations
Headline 3.8%** 0.4%
Body of the Press Release -2.30% 0.5%

*, **, *** indicate results significantly different from zero (H0=0) at the 0.10, 0.05, and 0.01 levels,
respectively, using two-tailed tests.

81

Table 10: Disclosure Salience for Firms Disclosing Negative Effect
Panel A of Table 10 provides descriptive statistics for the disclosure prominence of the hurricane news in
the press release (8-K report). For these types of firms there is no difference whether they disclose the “bad
news” that they are affected in the Headline/First paragraph or the body of the 8-K press release. Panel B
shows the disclosure salience for firms by earnings surprise. Firms are more likely to bundle negative
earnings surprise and the news that they are negatively affected in the body of the press release. Panel C
reports the disclosure salience for firms and earnings surprise returns. Stock The market reaction is negative
and significantly different from zero cumulative abnormal Fama-French three factor adjusted return.

Panel A: Disclosure of “Negative Hurricane Damage” in the Firm Press Release

Disclosure Salience Number Percent of Firms
Headline or First Paragraph
110 50%
Body of the Press Release
109 51%
Total
219 100%

Panel B: Disclosure of “Negative Hurricane Damage” and Whether A Firm Beats or Misses
Analysts’ Earnings Expectations

Disclosure of Damage
Beats
Expectations
Percent Misses
Expectations
Percent
Headline or First Paragraph 59 54% 51 47%
Body of the Press Release 51 46% 58 53%
Total 110 100% 109 100%

Panel C: Disclosure of “Negative Hurricane Damage” and Stock Market Returns
Disclosure of Damage Beats Expectations Misses Expectations
Headline or First Paragraph -1.01% -3.13%**
Body of the Press Release -0.1% -2.3%**

*, **, *** indicate results significantly different from zero (H0=0) at the 0.10, 0.05, and 0.01 levels,
respectively, using two-tailed tests.

82

Table 10: Disclosure Salience for Firms Disclosing Negative Effect (cont.)

Panel D of Table 10 provides descriptive statistics for the mean and median dollar damage for the 126 firms
that report dollar damage loss. Firms are more likely to disclose larger damage in the more prominent part
of the press release. The market reaction, however, is not dependent on the disclosure prominence. Panel
E shows the disclosure salience for firms and the cumulative abnormal returns. Panel E illustrates a more
attenuated market reaction form firms disclosing hurricane news in the body of the press release.

Panel D: Disclosure of Mean, Median Damage, and Cumulative Abnormal Returns for 126
firms Reporting Dollar Loss from Hurricanes

Disclosure of Damage
Number
of firms
Mean
Damage/
Total Assets
Median
Damage/
Total Assets CAR
Headline or First Paragraph
60
0.370 0.197 -.010
Body of the Press Release
66
0.163 0.087 -.012
Difference
6
0.20** 0.110*** .002
T-stat

(2.8) (3.12) (0.149)

Panel E: Disclosure of “Negative Hurricane Damage” and Stock Market Returns for 128 firms
that do not have Dollar Loss from Hurricanes

Disclosure of Damage
Number
of firms CAR
Headline or First Paragraph
73
-.023***
Body of the Press Release
55
-.011

*, **, *** indicate results significantly different from zero (H0=0) at the 0.10, 0.05, and 0.01 levels,
respectively, using two-tailed tests.

Abstract (if available)

Abstract This paper investigates the disclosure practices of firms affected by hurricanes. I document that when a hurricane hits, there is an increase in investor uncertainty. During the hurricane period (approximately ten days), there is an increase in abnormal volume, stock volatility, spread, and illiquidity for firms that later report that they experienced hurricane damage. I find that firms with little to no impact from the hurricane disclose this information immediately after the hurricane. In contrast, firms impacted by the hurricane delay reporting the damage until the next earnings announcement. Furthermore, firms with “good news” that the hurricane had little damage to operations disclose this news in the headlines of the earnings release (high salience) while firms that disclose a negative impact are more likely to bury the news in the body of the earnings press release (low salience). I also find that hiding the news in the body of the text has attenuating effect (weaker stock market reaction) on those firms that disclose qualitative and not quantitative hurricane damage. The results suggest that more timely updates on hurricane damage to investors reduces stock price volatility.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Disclosure distance and earnings announcement returns

PDF

Expectation dynamics and stock returns

PDF

Insider editing on Wikipedia

PDF

Essays on the firm and stakeholders relationships: evidence from mergers & acquisitions and labor negotiations

PDF

Implied equity duration: international evidence

PDF

Private disclosure contracts within supply chains and managerial learning from stock prices

PDF

Investment behavior of mutual fund managers

PDF

Voluntary disclosure responses to mandated disclosure: evidence from Australian corporate tax transparency

PDF

Essays on information and financial economics

PDF

Executive compensation: the trend toward one size fits all

PDF

Essays on the effect of cognitive constraints on financial decision-making

PDF

The changing policy environment and banks' financial decisions

PDF

Shareholder litigation as a disciplining device: evidence from firms' financial policies

PDF

Social movements and access to credit

PDF

Picky customers and expected returns

PDF

Changing fundamental analysis in the new economy: the case of DuPont analysis and STEM firms

PDF

Effectiveness of the SEC’s comment letters in initial public offerings

PDF

“What you see is all there is”: The effects of media co‐coverage on investors’ peer selection

PDF

Automated contracts and the lawyers who don't review them: adoption and use of machine learning technology

PDF

Does differential sensitivity to aggregate earnings shocks drive post-earnings-announcement drift?

Asset Metadata

Creator Stamenov, Ventsislav (author)

Core Title Understanding the disclosure practices of firms affected by a natural disaster: the case of hurricanes

School Marshall School of Business

Degree Doctor of Philosophy

Degree Program Business Administration

Degree Conferral Date 2021-08

Publication Date 08/13/2023

Defense Date 05/11/2021

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag mandatory disclosure,OAI-PMH Harvest,voluntary disclosure

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Dechow, Patricia (committee chair), Hoberg, Gerard (committee member), Sloan, Richard (committee member), Wong, Forester (committee member)

Creator Email stamenov@marshall.usc.edu,ventsi.stamenov@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC15723561

Unique identifier UC15723561

Legacy Identifier etd-StamenovVe-10039

Document Type Dissertation

Rights Stamenov, Ventsislav

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email cisadmin@lib.usc.edu