impact of global financial crisis on...
TRANSCRIPT
IMPACT OF GLOBAL FINANCIAL CRISIS ON UNITED
STATES STOCK MARKET PERFORMANCE
10 SECTORS AND FINANCIAL INDUSTRIES
ANALYSES
By
Kumari Ranjeeni
A thesis submitted in fulfillment of the requirements for a degree of
Master of Commerce
Copyright © 2014 by Kumari Ranjeeni
The University of the South Pacific
Faculty of Business and Economics
School of Accounting and Finance
November, 2014
DECLARATION OF ORIGINALITY
Statement by Author
I, Kumari Ranjeeni, declare that this thesis is my original piece of work and has been
compiled with my sole effort and hard work. All sources of information have been
properly referenced and assistance of any sort has been acknowledged.
Name : Kumari Ranjeeni Date: 13/11/2014
Student ID No: S11048124 Signature:
Statement by Supervisor
To my knowledge, this thesis has been compiled with the sole effort of Ms. Kumari
Ranjeeni under my supervision.
Name: Date: ________________
Signature: ___________________________
Designation: ______________________________
DEDICATION
I dedicate this thesis to my Mum, Mrs. Prakash Wati for being the source of my inspiration and also to my Daddy, Mr. Prakash Chandra for his love and wisdom.
iv
ACKNOWLEDGEMENT
I am extremely grateful to God for showering his blessing on me and giving me the
strength and wisdom to complete this thesis, thank you God. I am thankful to my
previous supervisor, Dr. Rohit Kishore, for his words of advice, encouragement and
guidance during the journey of my masters and also for having faith in me. I would also
like to extend my appreciation to Professor Biman Chand Prasad and the Head of
School, Professor Arvind Patel for being very considerate, not overloading me with
extra hours of teaching and also for encouraging my research work.
My sincere gratitude to my entire family members for encouraging my academic
potentials and for being very understanding, supportive and allowing me to devote
maximum time to research work. In particular, I would like to show my appreciation to
my daddy Mr. Prakash Chandra for his encouragement, love and wisdom, my dear mum
Mrs. Prakash Wati for being the source of my inspiration and for all her hard work and
sacrifices, my sister Ms. Ranjita Kumari for all her support and my brother Dr. Rohitash
Chandra for encouraging my academic potentials and guidance.
My heartiest thank to my brother, Dr. Rohitash Chandra for providing me high ranked
journal papers which were not available locally. These papers not only added value to
my literature review but simultaneously broadened my understanding and know – how
on the format of quality research presentation.
Special thanks to Alfred Deakin Professor Paresh Kumar Narayan (Deakin University)
and the University of the South Pacific for running a workshop and seminar on
econometric modeling which was very educational and assisted me in the development
of my research idea. Also, many thanks to Alfred Deakin Professor Paresh Kumar
Narayan for his valuable advices and overwhelming guidance.
v
ABSTRACT
This thesis investigates the impact of the news announcement of the Lehman Brothers’
(LB) bankruptcy on the performance of New York Stock Exchange (NYSE) sectors and
financial industries. Based on descriptive index level results, Bartram and Bodnar (2009)
conclude that the reaction of all sectors and industries was homogeneous during the LB
bankruptcy and equity investors could not benefit from diversification. Motivated by
Narayan and Sharma (2011) findings on firm and sector heterogeneity, this study
employs an event study approach to further examine the performance of sectors and
financial industries during the bankruptcy period. Daily data for a total of 488 firms is
examined. The main contribution of this thesis is that it provides evidence that sectors
behave heterogeneously during a stock market crisis and the significant adverse impact
from the LB bankruptcy is discriminatory towards the financial sector and the
diversified financial industry, which were most exposed to LB. This holds implications
for investors to devise profitable trading strategies.
This study unravels two new findings during LB bankruptcy. (1) Contrary to Bartram
and Bodnar (2009), this thesis provides evidence that the significant adverse impact
from the LB bankruptcy was not generalized to all sectors and financial industries. (2)
This study provides evidence that amongst the 10 sectors and 4 financial industries,
significant adverse impact from the LB bankruptcy was discriminatory towards the
financial sector and the diversified financial Industry. The results also provide evidence
in support of the US having a semi strong form of market efficiency. The robustness test
results support the main findings of the event study results.
vi
ABBREVIATIONS
AAR Average Abnormal Return
APT Arbitrage Pricing Theory
ARCH Auto Regressive Conditional Heteroskedasticity
Bnk Banks
CAAR Cumulative Average Abnormal Return
CAPM Capital Asset Pricing Model
CDO Collateralized Debt Obligations
CEE Central and Eastern European
CD Consumer Discretionary
CRSP Centre for Research in Security Prices
CS Consumer Staples
DF Diversified Financials
EM Emerging Markets
Eng Energy
Fin Financial
GARCH Generalized Auto Regressive Conditional Heteroskedasticity
GARCH – M Generalized Auto Regressive Conditional Heteroskedasticity in
Mean
GFC Global Financial Crisis
HC Health Care
Ind Industrial
Ins Insurance
ISE Istanbul Stock Exchange
IT Information Technology
LB Lehman Brothers
LCFIs Large Complex Financial Institutions
Mat Materials
MM Market Model
NYSE New York Stock Exchange
vii
OLS Ordinary Least Squares
REIT Real Estate Investment Trust
S&P 500 Standard & Poor’s 500
TMT Technology, Media and Telecommunication
TS Telecommunication Services
t – stats t – statistics
US United States of America
Uti Utilities
UK United Kingdom
viii
Table of Contents ABSTRACT ................................................................................................................................... v ABBREVIATIONS ...................................................................................................................... vi LIST OF TABLES ......................................................................................................................... x LIST OF FIGURES ........................................................................................................................ x CHAPTER 1: OVERVIEW OF THE RESEARCH .................................................................... 11
1.0 INTRODUCTION ....................................................................................................... 11 1.1 THE RELATIONSHIP BETWEEN LEHMAN BROTHERS BANKRUPTCY AND THE GLOBAL FINANCIAL CRISIS ..................................................................................... 13 1.2 RISK – RETURN RELATIONSHIP DURING STOCK MARKET CRASHES ........ 16 1.3 RESEARCH ISSUES AND CONTRIBUTION OF THE STUDY ............................. 17 1.4 AIM AND OBJECTIVES ............................................................................................ 19 1.5 OUTLINE OF THIS THESIS ...................................................................................... 21 1.6 CONCLUSION ............................................................................................................ 22
CHAPTER 2: LITERATURE REVIEW ..................................................................................... 23 2.0 INTRODUCTION ....................................................................................................... 23 2.1 FINANCIAL LIBERALIZATION .............................................................................. 23 2.2 THE US STOCK MARKET: ITS SIGNIFICANCE AND VOLATILITY STRUCTURE DURING STOCK MARKET CRASHES ....................................................... 26 2.3 THEORIES ON THE RISK AND RETURN RELATIONSHIP ................................ 27
2.3.1 ASSET PRICING THEORY ............................................................................... 27 2.3.2 STOCK MARKET SHOCKS AND CONFLICTING THEORIES ON VOLATILITY ASYMMETRY ........................................................................................... 29
2.4 AN OVERVIEW OF THE IMPACT OF STOCK MARKET CRASHES .................. 33 2.4.1 CONTAGION EFFECTS FROM THE ASIAN FINANCIAL CRISIS .............. 36 2.4.1.1 THE IMPACT OF THE ASIAN FINANCIAL CRISIS ON STOCK MARKET PERFORMANCE ................................................................................................................. 36 2.4.1.2 CONTAGION EFFECTS ..................................................................................... 37 2.4.2 CONTAGION EFFECTS FROM THE GFC AND LB BANKRUPTCY ........... 38 2.4.2.1 THE IMPACT OF THE GFC AND LB BANKRUPTCY ON STOCK MARKET PERFORMANCE ................................................................................................................. 38 2.4.2.2 CONTAGION EFFECTS ..................................................................................... 39 2.4.3 CONTAGION EFFECTS, GLOBAL AND LOCAL RISK FACTORS ............. 42
2.5 CONCLUSION ............................................................................................................ 43 CHAPTER 3: DATA, METHODOLOGY AND HYPOTHESES DEVELOPMENT ............... 44
3.0 INTRODUCTION ....................................................................................................... 44 3.1 DATA AND SAMPLE SELECTION ......................................................................... 45 3.2 EVENT STUDY .......................................................................................................... 49 3.3 ROBUSTNESS TESTS FOR THE EVENT STUDY RESULTS ............................... 64 3.4 HYPOTHESIS DEVELOPMENT ............................................................................... 64
3.4.1 LB BANKRUPTCY AND THE US FINANCIAL SECTOR .............................................. 65
ix
3.4.2 LB BANKRUPTCY AND THE FINANCIAL INDUSTRIES ............................................. 68 3.5 CONCLUSION ............................................................................................................ 73
CHAPTER 4: DESCRIPTIVE STATISTICS RESULTS ........................................................... 74 4.0 INTRODUCTION ....................................................................................................... 74 4.1 DESCRIPTIVE STATISTICS FOR THE SAMPLE USED IN THE STANDARD OLS MARKET MODEL ......................................................................................................... 74 4.2 DESCRIPTIVE STATISTICS FOR THE VARIANT OF THE STANDARD OLS MARKET MODEL ................................................................................................................. 77 4.3 CONCLUSION .................................................................................................................. 79
CHAPTER 5: EVENT STUDY RESULTS FOR 10 SECTORS BASED ON THE STANDARD OLS MM ...................................................................................................................................... 80
5.0 INTRODUCTION ....................................................................................................... 80 5.1 RESULTS FOR 10 SECTORS BASED ON THE STANDARD MM ........................ 81 5.2 CONCLUSION ............................................................................................................ 88
CHAPTER 6: EVENT STUDY RESULTS FOR THE 10 SECTORS ........................................ 89 6.0 INTRODUCTION ....................................................................................................... 89 6.1 OLS MM RESULTS BASED ON AN ESTIMATION PERIOD OF 150 DAYS BEFORE 30 JUNE 2007 ......................................................................................................... 90 6.2 ROBUSTNESS TEST RESULTS BASED ON THE MARKET ADJUSTED RETURN MODEL ..................................................................................................................................... 94 6.3 CONCLUSION ............................................................................................................ 97
CHAPTER 7: EVENT STUDY RESULTS FOR THE FINANCIAL INDUSTRIES ................ 98 7.0 INTRODUCTION ....................................................................................................... 98 7.1 THE OLS MM RESULTS BASED ON AN ESTIMATION PERIOD OF 150 DAYS BEFORE 30 JUNE 2007 ......................................................................................................... 99 7.2 ROBUSTNESS TEST RESULTS BASED ON THE MARKET ADJUSTED RETURN MODEL ................................................................................................................ 102 7.3 CONCLUSION .......................................................................................................... 104
CHAPTER 8: CONCLUSION.................................................................................................. 105 Implications and Future Research .............................................................................................. 105
8.0 INTRODUCTION ..................................................................................................... 105 8.1 MAIN FINDINGS ..................................................................................................... 106 8.2 IMPLICATIONS ....................................................................................................... 107 8.3 LIMITATIONS AND SUGGESTIONS FOR FUTURE RESEARCH ..................... 110
REFERENCE ............................................................................................................................. 111 APPENDIX A ............................................................................................................................ 118 APPENDIX B ............................................................................................................................ 119
x
LIST OF TABLES
Table 3.1: Sample Data Set .............................................................................................. 48 Table 3.2: Correlation Results Between LB and each of the 10 sectors .......................... 68 Table 4.1: Descriptive statistics for the sample used in the standard OLS MM .............. 76 Table 4.2: Descriptive statistics for the sample used in the variant of the standard OLS MM ................................................................................................................................... 78 Table 5.1: Findings of t-tests on the analysis of daily AARs per sector based on the standard OLS MM ........................................................................................................... 83 Table 5.2 Findings of t-tests on the analysis of CAARs per sector based on the standard OLS MM .......................................................................................................................... 87 Table 6.1: Findings of t-tests on the analysis of daily AARs per sector based on the variant of the standard OLS MM ..................................................................................... 93 Table 6.2: Findings of t-tests on the analysis of CAARs per sector based on the variant of the standard OLS MM ................................................................................................. 93 Table 6.3: Significance test between the Financial sector and each of the other nine sectors ............................................................................................................................... 94 Table 6.4: Findings of t-tests on the analysis of daily AARs per sector based on the market adjusted return model ........................................................................................... 96 Table 6.5: Findings of t-tests on the analysis of CAARs per sector based on the market adjusted return model ....................................................................................................... 96 Table 7.1: Findings of t-tests on the analysis of daily AARs per financial industry based on the variant of the standard OLS MM ........................................................................ 101 Table 7.2: Findings of t-tests on the analysis of CAARs per financial industry based on the variant of the standard OLS MM ............................................................................. 101 Table 7.3: Findings of t-tests on the analysis of daily AARs per financial industry based on the market adjusted return model .............................................................................. 103 Table 7.4: Findings of t-tests on the analysis of CAARs per financial industry based on the market adjusted return model ................................................................................... 103
LIST OF FIGURES
Figure 1.1: Illustration of the S&P 500 Index Composition ............................................ 13 Figure 1.2: Illustration of Research Issues ....................................................................... 20
11
CHAPTER 1: OVERVIEW OF THE RESEARCH
1.0 INTRODUCTION
The non-strategic bankruptcy of Lehman Brothers’ (LB)1 occurred in the United States
of America (US) financial sector. The market considers a non-strategic bankruptcy to be
a bad news having an adverse impact on the stock market performance (Coelho and
Taffler 2008). In addition, a crisis originating from the financial sector is widely
believed to be more contagious and riskier than those from other sectors of the
economy 2 (Kaufman 1994). Evidence suggests that the LB bankruptcy triggered
incredible declines in index levels and escalated price volatility universally (Bartram and
Bodnar 2009; Chong 2011; Eichler et al. 2011; Samarakoon 2011). The index levels at
aggregate level proxies the behaviour of the stock market firms and sectors. A
disadvantage of examining index level performance is that all firms and sector
constituents of the index are erroneously assumed to be homogeneous. A recent study by
Narayan and Sharma (2011) suggest that share market firms and sectors are
heterogeneous in nature, with an implication that the LB bankruptcy may not impose
significant adverse impact across all sectors. The question that then arises is: Did the LB
bankruptcy have a homogenous impact on the performance of all sectors or was the
adverse impact heterogeneous towards the financial sector that was most exposed to LB?
To my knowledge, this research question has not been addressed so far. Therefore, this
study takes the initiative of using a disaggregated approach to examine the US stock
market performance at sector and financial industry level around the event of LB
bankruptcy.
This study investigates whether the financial sector, identified as the epicenter of the
Global Financial Crisis (GFC), and the diversified financial industry were the most
1 LB was a distressed financial institution that filed for Chapter 11 bankruptcy with petition no. 308-13555 on 15 September 2008 (Chong 2011; Dumontaux and Pop 2009). 2 As the saying goes “one man’s (sectors) crisis is another man’s (sector’s) road to prosperity” (Gorban et al. 2010).
12
significantly adversely affected during LB bankruptcy. Overall the findings of this study
will hold economic implications towards appropriate investor strategies to be used in
times of a crisis. How? If the results support all the propositions then that would imply
that sector heterogeneity exists even in times of a crisis. Therefore, investors could
benefit by devising strategies that would hedge against severe losses or provide optimal
returns. Otherwise, if the findings reveal that all sectors were generally significantly
affected then that would support the descriptive index level findings of Bartram and
Bodnar (2009) and imply that diversification benefits did not exist during GFC (Bartram
and Bodnar 2009).
In this study, firms are distributed into sectors and financial industries according to their
respective Global Industry Classification Standards (GICS). An “industry” specifically
refers to a group of firms involved in same lines of business operations while a “sector”
comprises of a group of similar industries. Accordingly, there are 10 sectors, namely;
Consumer Discretionary (CD), Consumer Staples (CS), Energy (Eng), Financial (Fin),
Health Care (HC), Industrials (Ind), Information Technology (IT), Materials (Mat),
Utilities (Uti) and Telecommunication services (TS). The four financial industries are
namely, Banks (Bnk), Diversified Financials (DF), Insurance (Ins) and Real Estate
Investment Trust (REIT). In this study, the words “banks” and “banking industry” are
used interchangeably. Figure 1.1 provides an illustration of the disaggregation of the
Standard & Poor’s 500 (S&P 500) index, which is used for investigation in this study,
into 10 sectors and 4 financial industries.
The rest of this chapter is organized as follows. In Section 1.1, the relationship between
LB bankruptcy and the GFC is discussed. In Section 1.2, the risk and return relationship
during stock market crashes is discussed. In Section 1.3, the main research issues
motivating this thesis along with the contribution of this study are discussed. In Section
1.4, the aim and specific objectives of this thesis are provided. Finally, an outline of this
thesis and chapter conclusion is provided in Sections 1.5 and 1.6 respectively.
13
Figure 1.1: Illustration of the S&P 500 Index Composition
1.1 THE RELATIONSHIP BETWEEN LEHMAN BROTHERS
BANKRUPTCY AND THE GLOBAL FINANCIAL CRISIS
The 150 years old LB was the fourth largest investment bank in the US and was
considered one of the Wall Street's biggest dealers in fixed-interest trading (Dumontaux
and Pop 2009). LB operations involved dealings with other banks, pension funds and
hedge funds. Next, a brief overview of the factors contributing to the GFC is provided
followed by discussions on how LB bankruptcy triggered the GFC.
In the past three decades, there has been a shift in paradigm towards a capitalist
economy. The shift in paradigm was supported by the governments globally to avoid
the history of deflation and depression of the 1930s (caused by the government
intervention policies) subsequent to the Wall Street crash in 1929 from repeating itself.
In support, the classical economists believed that the economy would work best if left on
its own without government intervention. The competitive market forces of demand and
Sectors
S & P 500 Index
CD CS HC Fin Eng IT TSUti Mat Ind
Fin Industries Bnk DF Ins REIT
14
supply was perceived to bring about efficiency in an economy via enhancement in the
rate of economic growth and living standards. Accordingly, informed monetary policy
over the activist fiscal measures was accepted as the tool for preserving the liberalized
world from the old cycle of boom and bust. On the contrary, the paradigm shift resulted
in the GFC having the steepest economic turmoil in nearly 80 years since the great
depression of the 1930s.
The GFC originated in the US from the end of the US house price bubble and the
subsequent collapse of the subprime mortgages. Although the failure in the subprime
mortgage market had already started affecting the US stock market, its performance was
most significantly influenced by LB bankruptcy. LB was at the core of the financial
system. It was the fourth largest investment bank in the US and was considered one of
Wall Street's biggest dealers in fixed-interest trading. LB operations involved dealings
with other banks, pension funds and firms like hedge funds. Consequently, its dealings
not only affected the US residents but also indirectly affected millions of people
globally.
Furthermore, LB had heavily invested in securities linked to the US sub-prime mortgage
market3. Therefore, delinquent mortgage payments by subprime borrowers resulted in
LB incurring severe losses. However, unlike many other large distressed financial
institutions (Bear Sterns, Fannie Mae, Freddy Mac, American Insurance Group and
Citigroup) which were also at the core of the US financial system, LB was not bailed out
by the government in the absence of feasible private sector solution. The bailed out
financial institutions were considered systematically important or “too big to fail” and
hence were rescued by pumping huge amounts of tax payers’ money (Dumontaux and
Pop 2009).
3 BBC News Channel, “Q&A: Lehman Brothers bank collapse”, 16 September 2008, available at: http://news.bbc.co.uk/1/hi/business/7615974.stm.
15
Serious concerns were raised by many observers, economists and academia’s for not
rescuing LB. Critics argue that if Bear Sterns was “too large and important and too
inextricably enmeshed in the global financial system to allow it to fail, so should
Lehman have been” (Bartholomeusz 2008). The government’s decision for not bailing
out LB was as follows:
“…unlike in the case of Bear Sterns, market participants have had sufficient
time to prepare themselves to absorb the collateral damages eventually caused
by the imminent collapse of Lehman. Moreover, in contrast to Bear Sterns,
Lehman had direct access to short-term facilities (PDCF) from the Federal
Reserve. Top government officials also pointed out that they viewed Fannie Mae
and Freddie Mac as far more systemically important than Lehman because the
two mortgage giants own or guarantee about half of home loans originated in
the US” (Dumontaux and Pop 2009).
LB bankruptcy was announced on 15 September, 2008 (Chong 2011; Dumontaux and
Pop 2009). It had $639 billion worth of total assets, hence, making it the biggest collapse
in the US history (Dumontaux and Pop 2009). LB subsequent liquidation signaled
defects in the loan securitization process and mortgage linked securities. Investor
confidence in the securities market was greatly affected and investors started to sell their
shares. There was an increase in the supply of shares in the market which led to a fall in
share price. As a result of LB bankruptcy:
“the specter of systemic risk raised widespread fears of a full-scale collapse of
the US financial sector due to financial contagion and concerns about significant
disturbances outside the US, in international financial markets” (Dumontaux
and Pop 2009).
Since LB was a core member of the financial system, LB bankruptcy drastically affected
the Wall Street, the economy and all investors. Globalization and rapid transmission of
information across financial markets resulted in financial contagion and caused the crisis
to spread from one country to another irrespective of each country’s economic
fundamental. Financial contagion was demonstrated by stock market crashes around the
16
world following the collapse of the US stock market. Thus, the liquidation of LB is
believed to have marked the beginning of the anticlimax era in the share market
performance globally.
1.2 RISK – RETURN RELATIONSHIP DURING STOCK MARKET CRASHES
Numerous researches have analyzed the risk and return relationship during periods of
high information asymmetry and unstable market conditions (Glosten et al. 1993; Fang
2001; Celikkol et al. 2010). In times of a shock, empirical findings on the relationship
between expected returns and conditional volatility are diverse. Previously, time series
models assumed that both positive and negative return shocks symmetrically increased
volatility (Wu 2001; Ederington and Guan 2010). However, with the use of more
flexible estimation models (Nelson 1991; Glosten et al. 1993), asymmetry in volatility
was found particularly in the US stock market (Black 1976; Christie 1982; Ederington
and Guan 2010).
Furthermore, a negative return shock was found triggering higher volatility (greater
impact) in comparison to positive return shocks of the same magnitude (Ederington and
Guan 2010). Many studies have argued that a volatility shock could significantly
increase risk premium resulting in a positive relationship between risk and return
because investors would require greater expected return from riskier securities in order
to be compensated for taking on the additional risk. Conversely, other studies have
found that investors may not necessarily require higher risk premiums during risky
periods which could result in a negative relationship between risk and return (Glosten et
al. 1993). Accordingly, the risk - return relationship across time is controversial and
yields conflicting results.
17
1.3 RESEARCH ISSUES AND CONTRIBUTION OF THE STUDY
This study is motivated by three pioneering studies; namely, Bartram and Bodnar
(2009), Pichardo and Bacon (2009) and Dumontaux and Pop (2009). Bartram and
Bodnar (2009) investigated the impact of LB bankruptcy on the US stock market
performance for the period 12 September 2008 to 27 October 2008. The findings
revealed that all industries and sectors index returns were adversely affected during LB
bankruptcy. Since share market sectors are found to be integrated, shock from any sector
may influence other sector returns (Wang et al. 2005). Accordingly, the decline in index
returns from LB bankruptcy as found by Bartram and Bodnar (2009) was expected.
However, Bartram and Bodnar (2009) provide only descriptive index level results. A
recent study by Narayan and Sharma (2011) suggest that share market firms and sectors
are heterogeneous in nature, with an implication that the LB bankruptcy may not impose
significant adverse impact across all sectors. Therefore, conclusions cannot be drawn on
the significance (if any) of the impact of LB bankruptcy on the sectors and industries.
Although Pichardo and Bacon (2009) and Dumontaux and Pop (2009) have used
significance tests to examine abnormal return performance during LB bankruptcy, the
investigation is conducted only for “large” 4 financial firms operating in the same
industry as LB or operating mainly in other fields of finance. Pichardo and Bacon (2009)
examine abnormal returns of 15 investment firms out of which 9 had significant stake in
LB for the period 1 September 2008 to 27 October 2008. Dumontaux and Pop (2009)
examine the contagion effects from LB bankruptcy on the large US surviving financial
institutions for the year 2008. Therefore, Pichardo and Bacon (2009) and Dumontaux
and Pop (2009) have examined the impact of LB only on large financial firms and hence
the results could not be generalized to all financial industries. This study extends the
extant literature by thoroughly investigating the impact of LB bankruptcy on NYSE
sectors and financial industries.
4 “Large” is defined as firms “that reported total assets higher than US$ 1 billion in the last audited financial report before the event date” (Dumontaux and Pop 2009).
18
In light of the idea of firm and sector heterogeneity (Narayan and Sharma 2011), this
study examines whether (1) the financial sector and (2) the diversified financial industry
were most significantly adversely affected during LB bankruptcy. These are the main
research questions of this study and are motivated by the following reasons. Acharya et
al. (2009) mention LB bankruptcy having significant systematic risk and Bartram and
Bodnar (2009) find increased correlation across the US sectors following LB
bankruptcy. However, the extant literature also suggest that the firms having direct
credit or investment exposures to LB suffered more than those not having direct
exposure to LB (Pichardo and Bacon 2009; Chakrabarty and Zhang 2010; Dumontaux
and Pop 2009). This is an indication of LB bankruptcy having a heterogeneous effect on
the performance of the US firms, depending on their exposure to LB. This indicates that
LB bankruptcy had a heterogeneous effect on the US firm performance depending on the
firm’s exposure to LB. Presumably, those sectors and industries highly
correlated/associated with LB would have been exposed to higher risks (volatility) and
vice versa. LB was at the core of the US financial system with high investments in
subprime mortgages. LB, being an investment bank and a constituent of the diversified
financial industry, would have been more closely associated with other diversified
financial firms than firms from the other three financial industries (banks, insurances and
REITs). This motivates the investigation of whether or not LB bankruptcy most
significantly adversely affected the performance of the US financial sector and the
diversified financial industry.
The main contributions of this study are discussed as follows. First, it provides evidence
that sectors behave heterogeneously during a crisis and unravels that the significant
adverse impact from the LB bankruptcy is discriminatory towards the financial sector
and the diversified financial industry that were most exposed to LB. This holds
implications for investors to devise profitable trading strategies to hedge against severe
losses or provide optimal returns during a crisis. Second, this study conducts a detailed
analysis of NYSE performance at sector and financial industry levels during the LB
bankruptcy. All financial firms (irrespective of firm size) and financial industries
19
inclusive of REIT and insurance are examined. Furthermore, the impact of LB
bankruptcy on all the 10 sectors is examined. Consequently, stronger conclusions could
be drawn on the pervasiveness of the impact from LB bankruptcy. This will contribute
towards the debate on the government’s non-bailout decision for LB. Finally, the
empirical analysis used in this study is not only robust but also has strong theoretical
foundation5. The event study approach employed follows the theory of market efficiency
and supports Pichardo and Bacon (2009) on the efficiency of the US market during LB
bankruptcy.
Finally, the findings of this research are important as it investigates on the most
influential US equity market using the S&P 500 index composition data. Consequently,
it provides an understanding on the dynamics of the US equity market, which is not only
the largest but also the most prominent equity market, stock return and volatility. Also,
the S&P 500 index data is used as a proxy for the US stock market because LB was
initially listed on it and subsequently delisted following its bankruptcy. Therefore, it will
act as a better means to gauge the effect of LB bankruptcy on the US stock market.
1.4 AIM AND OBJECTIVES
The aim of this thesis is in twofold (see Figure 1.2 for illustration). Firstly, to
empirically test whether or not the financial sector was the most significantly adversely
affected by LB bankruptcy (1). Secondly, it is to examine whether or not the diversified
financial industry was the most significantly adversely affected by LB bankruptcy (2).
5 Conversely, Bartram and Bodnar (2009) findings were only descriptive.
20
Figure 1.2: Illustration of Research Issues
In order to achieve the purposes of this research the following objectives have been
articulated.
1. To establish the event window for event study purpose.
2. To use an event study approach and examine the impact of LB bankruptcy on the
performance of each of the 10 sectors.
3. To analyze whether or not the financial sector was the most significantly
adversely affected during LB bankruptcy.
4. To use an event study approach and examine the impact of LB bankruptcy on the
performance of each of the financial industries.
5. To analyze whether or not the diversified financial industry was the most
significantly adversely affected during LB bankruptcy.
ration of
(1) Financial Sector
(2) Diversified Financial Industry
LB bankruptcy
Impact of LB bankruptcy on (1)
and (2)?
21
6. To use the market adjusted return model to check the robustness of event study
results.
7. To draw implications from the findings and contribute to the extant literature.
1.5 OUTLINE OF THIS THESIS
This thesis is divided into 8 chapters. The current chapter, which is an introductory
chapter, is followed by the literature review in Chapter 2. Literature review begins with
discussions on financial liberalization resulting in the formation of the world equity
market followed by discussions on the dominant role played by the US in influencing
other stock market’s return and volatility. Next, discussion on the competing theories on
the risk and return relationship is provided. An overview of stock market crashes is also
provided.
Chapter 3 is on Research Design. This chapter provides a detailed description of the
sample used along with its associated advantages. Next, research methods used in this
thesis are discussed. It discusses on the event study approach with explanations on the
idea behind its usage followed by discussions on its common applications and methods
to conduct the event study. The standard OLS MM (with estimation period immediately
before the event period) and a variant of the standard OLS MM (with estimation period
before the GFC) employed in this thesis are discussed. Finally, the market adjusted
return model employed to check the robustness of the event study results is discussed.
Chapters 4, 5, 6 and 7 are on the findings of this thesis. Chapter 4 provides the
descriptive statistics for the sample used in this thesis. Chapter 5 presents event study
results for each of the 10 sectors based on the standard OLS MM. The standard OLS
MM has an estimation period of 150 days before the event window of [−30, +30] trading
days. Standardized cross sectional test is used to test the significance of the standard
OLS MM results.
22
Chapter 6 presents event study results for each of the 10 sectors based on the variant of
the standard OLS MM that has an estimation period of 150 days before the GFC (30
June 2007) and [−4, 0] trading days event window. Standardized cross sectional test is
used to test the significance of the variant of the standard OLS MM results. Next, results
for the robustness test conducted based on the market adjusted return model is presented.
Ordinary cross sectional test approach is used to test the significance of the market
adjusted return model results.
Chapter 7 presents event study results for each of the financial industries based on the
variant of the standard OLS MM that has an estimation period of 150 days before the
GFC (30 June 2007) and [−4, 0] trading days event window. Standardized cross
sectional test is used to test the significance of the variant of the standard OLS MM
results. Next, robustness test results based on the market adjusted return model are
presented. Ordinary cross sectional test approach is used to test the significance of the
market adjusted return model results.
Finally, Chapter 8 concludes this study with the main findings of this research and
economic implications of results. It also sheds light on this study’s research limitations
and provides direction for future research.
1.6 CONCLUSION
In this chapter, the research motivations are outlined and the research gap is identified
based on the extant literature. The chapter also highlights the research issues and
discusses the methods employed to address the research issues. The contributions
followed by the aim and objectives of this study are discussed. Moreover, a brief
discussion on LB, LB bankruptcy, GFC and the risk - return relationship during stock
market crashes is provided. In the next chapter, a detailed literature review is provided.
23
CHAPTER 2: LITERATURE REVIEW
2.0 INTRODUCTION
Over the past two decades, stock markets have undergone rapid changes and
developments. Financial liberalization is one such revolution that has interconnected
global equity markets. The US stock market plays a dominant role in the world equity
market.
In Section 2.1, financial liberalization and its impact on stock market interdependence
and volatility is discussed. In Section 2.2, the US stock market, its significance and
volatility structure during stock market crashes is discussed. In Section 2.3, competing
theories on the risk and return relationship are discussed. In Section 2.4, an overview of
stock market crashes is provided. Finally, chapter conclusion is given in Section 2.5.
2.1 FINANCIAL LIBERALIZATION
Financial liberalization has opened up domestic stock markets to enable trading of
shares at an international scale. De la Torre et al. (2007) explain stock market
liberalization as governments decision “to allow foreign investors to purchase shares in
the local stock market and domestic investors to purchase shares abroad”. Conversely, in
the early 1980s, there was an absence of legal method to enable foreign investment in
emerging (domestic) market equities (Kawakatsu and Morey 1999).
Furthermore, advocates of financial liberalization believe that it will result in better
allocation of resources, international risk sharing (Iwata and Wu 2009), and lower cost
of capital (Kim and Singal 2000; De la Torre et al. 2007), market efficiency (Kwan and
Reyes 1997; Kim and Singal 2000; Cuñado et al. 2006) and enhancement of economic
growth (Hale 1994).
24
Iwata and Wu (2009) examine major Latin American countries (Mexico, Brazil, Chile
and US) and find that liberalizing countries were able to achieve risk - sharing by
hedging against exogenous and idiosyncratic financial shocks. However, they also find
that in order to reap the full benefits of international risk sharing, “stock market
liberalization needs to be accompanied by other measures of economic integration (such
as international trade and migration)”.
Additionally, numerous studies have assessed the effect of stock market liberalization on
the cost of equity capital and provide evidence of reduced share price around the
liberalization date and subsequent decline in the cost of capital (De la Torre et al. 2007).
Kim and Singal (2000) also find reduced cost of capital after liberalization. Hale (1994)
identifies liberalization as a factor contributing to the growth in developing countries
stock market by changing the “attitudes toward both the role of securities markets as
allocators of capital and potential vehicle for attracting foreign investment”.
Furthermore, financial liberalization is perceived to enhance market efficiency via equity
pricing. According to the efficient market hypothesis, “as equity markets are liberalized
and made more open to the public, equity prices should reflect the increased availability
of information and be more efficiently priced” (Kawakatsu and Morey 1999). In the
contrary, Kawakatsu and Morey (1999) find that EM’s were already efficient prior to
actual liberalization and liberalization does not seem to have improved the efficiency.
Moreover, “according to finance literature, stock market volatility could either increase
or decrease when markets are opened” (Cuñado et al. 2006). Following liberalization,
stock markets are anticipated to become more informational efficient (Kwan and Reyes
1997; Kim and Singal 2000; Cuñado et al. 2006). Therefore, stock prices would quickly
comprehend and reflect relevant information. As a result, it is feared that stock market
volatility could increase. This has motivated researchers to investigate the impact of
25
financial liberalization on stock market return and volatility and the findings are
discussed below.
Financial liberalization thesis suggests that the volatility in stock markets will decline
following the opening up of stock markets to foreign investors (Kassimatis 2002)
because a given amount of risk will be shared amongst many investors (Law 2006). The
gradual development and diversification of the markets leading to lower volatility will
also reduce the market’s sensitivity to new information (Cuñado et al. 2006). In support
to the financial liberalization theory, Kwan and Reyes (1997), Kim and Singal (2000),
Kassimatis (2002), Cuñado et al. (2006), James and Karoglou (2010) find that following
liberalization, stock market volatility significantly declined in emerging stock
exchanges.
Conversely, Huang and Yang (2000) find stock market volatility increasing in three
EM’s (South Korea, Mexico and Turkey) and diminishing volatility for the other four
countries (Chile, Malaysia and the Philippines) after liberalization. Edwards et al. (2003)
find that unlike Asian countries, Latin American stock markets become less volatile after
liberalization. Aggarwal et al. (1999) find both increases and decreases in volatility
depending on the country and on the sequence of events.
Conclusively, there are mixed findings on the impact of financial liberalization on stock
market volatility. Nevertheless, it has contributed to the internationalization of finance
and capital market linkages. Therefore, financial markets have become highly
interdependent. Consequently, the chance of a shock from an influential economy being
transmitted to cross - border markets internationally has significantly increased. In the
next section, a brief discussion of the most influential equity market, the US stock
market, is provided.
26
2.2 THE US STOCK MARKET: ITS SIGNIFICANCE AND VOLATILITY
STRUCTURE DURING STOCK MARKET CRASHES
Globalization and financial liberalization interconnected the stock markets and together
they are represented by the world equity market. The US equity market, which is the
world’s largest (Ammann and Kessler 2008) and the most influential market, explains
other national stock markets variability of returns and volatility (Kearney 2000).
Furthermore, stock market crashes 6 further strengthened the relationship between
international stock markets. During stock market crashes, the US equity market plays a
dominant role (Chen et al. 2006) in influencing the Asian market (Arshanapalli et al.
1995), emerging and developed stock markets (Soydemir 2000) and Pacific-Basin
markets (Liu et al. 1998). Hon et al. (2007) find results supporting large economy
assumption that movements in foreign markets is explained by US stock returns,
however, the vice versa does not hold.
In times of a shock, aggregate stock market volatility in the US was particularly found to
be asymmetric in nature (Ederington and Guan 2010). “Black (1976) and Christie
(1982) were among the first to document and explain the asymmetric volatility property
of individual stock returns in the U.S equity markets” (Wu 2001). Section 2.3.2 will
shed light on volatility asymmetries during stock market crashes. Next, theories on the
risk and return relationship are discussed.
6 See Section 2.4 on how US equity market has influenced other equity markets return and volatility in times of stock market crashes.
27
2.3 THEORIES ON THE RISK AND RETURN RELATIONSHIP
2.3.1 ASSET PRICING THEORY
Investment in shares is classified as risky due to its returns variability and consequently
uncertainty associated with the expected return from materializing. Expected return
reflects the best estimate of the holding period7 return, relative to all possible outcomes,
that investors expect to receive. In other words, from the firms’ perspective, expected
return represents the cost of equity. Therefore, an understanding of the relationship
between risk and return is vital because it has implications for stock valuation8 (Parrino
and Kidwell 2009).
Furthermore, total risk comprises of unsystematic risk and systematic risk. Unsystematic
risk is also known as diversification risk and refers to firm specific or idiosyncratic risk.
Such types of risk can easily be eliminated by diversifying investments across and
within less than perfectly positively correlated asset classes. On the other hand,
systematic risk refers to market inherent risk and is non diversifiable. Beta is used to
measure systematic risk (Beal and McKeown 2009; Parrino and Kidwell 2009).
Therefore, investors demand a premium relative to beta for undertaking systematic risk.
Accordingly, the capital asset pricing model (CAPM) was developed by Sharpe (1964)
and Lintner (1965) to price an individual security or a portfolio relative to its systematic
risk9. This gave birth to the asset pricing theory for which Sharpe won a Nobel Prize in
1990 (Fama and French 2004). The CAPM model is based on several assumptions
(Head 2008). However, in this study, assumptions of rational investor behavior,
diversification and efficient market hypothesis are particularly explored. The CAPM 7 A standardized holding period of one year is usually assumed by CAPM to enable comparison of different securities returns (Head 2008). 8 “The time value of money is directly related to the returns that investors require” (Parrino and Kidwell 2009). 9 According to the finance theory, investors are rewarded for only taking systematic risk because unsystematic risk is diversifiable.
28
assumption on rational investor behavior states that investors are risk averse and will
only undertake additional risk if they are compensated for it. Parrino and Kidwell (2009)
state that higher rate of return required by investors from riskier assets is “one of the
fundamental relations in finance”.
Additionally, CAPM assumes that rational investors will hold diversified portfolios. As
a result, unsystematic risk will be eliminated and they will only be exposed to systematic
risk10. Accordingly, diversified investors will face less risk. “Because they face less
risk, the diversified investors will be willing to pay higher prices for assets than the other
investors. Therefore, an individual asset’s expected returns will be lower than its total
risk (systematic plus unsystematic risk). Diversified investors will bid up prices for
assets to the point at which they are just being compensated for the systematic risks they
must bear” (Parrino and Kidwell 2009). In other words, “only systematic risk affects
expected returns on investment” (Parrino and Kidwell 2009). Accordingly, CAPM
compensates risk averse investors for undertaking systematic risk11 by establishing a
relationship between systematic risk (market risk) and return.
Additionally, the market efficiency hypothesis is a “theory concerning the extent to
which information is reflected in security prices and how information is incorporated
into security prices” (Parrino and Kidwell 2009). CAPM assumes “that perfect
information is freely available to all investors”. In other words, according to CAPM,
“security prices reflect all available information”, thus, there is a strong form of market
efficiency (Parrino and Kidwell 2009). However, “it is widely accepted that insiders
have information that is not reflected in the security prices. Therefore, the concept of a
strong form market efficiency represents the ideal case rather than the real world”,
(Parrino and Kidwell 2009). Nevertheless, developed countries such as the US has semi-
10 Conversely, investors without well diversified portfolios will be exposed to both systematic and unsystematic risk (Parrino and Kidwell 2009). 11 Share’s systematic risk is measured by the variability of share’s returns with respect to the market’s return (Beal and McKeown 2009).
29
strong form of market efficiency. “In a market characterized by this sort of efficiency, as
soon as information becomes public, it is quickly reflected in stock prices through
trading activity” (Parrino and Kidwell 2009). Therefore, in this thesis, LB bankruptcy
will be investigated using the event study approach12.
Moreover, the price of equity market risk is measured by the risk premium. The CAPM
beta13 (β), which is used as a proxy to ascertain the responsiveness of the risk of a share
to the total market risk, enables the computation of expected return for that particular
share. In any particular market, the risk free rate and risk premium are the same and the
difference in expected return of shares is attributed to their respective betas (Beal and
McKeown 2009). Conclusively, “asset pricing models imply a positive relationship
between risk and return under the assumption of investor risk aversion” (Xing and Howe
2003).
2.3.2 STOCK MARKET SHOCKS AND CONFLICTING THEORIES ON
VOLATILITY ASYMMETRY
Stock market is affected by negative as well as positive shocks. In times of a shock,
empirical findings on the relationship between expected returns and conditional
volatility are diverse. In the earlier time series models [Auto Regressive Conditional
Heteroskedasticity (ARCH) and Generalized Auto Regressive Conditional
Heteroskedasticity (GARCH)] it was assumed that volatility symmetrically increases
subsequent to positive as well as negative return shocks (Wu 2001; Ederington and Guan
2010). However, with the use of more flexible estimation models [asymmetry ARCH
models of Nelson (1991) and Glosten et al. (1993)], asymmetry in volatility was found
(Wu 2001; Ederington and Guan 2010) and such models significantly outperformed
those which did not capture asymmetric volatility behavior (Wu 2001). In equity
12 Refer to chapter 3 for further justifications for the use of an event study approach. 13 The market has a beta of 1
30
markets, asymmetric volatility implies negative relationship14 between contemporaneous
return and next period’s conditional return variance (Wu 2001). Furthermore, a negative
return shock triggers relatively higher volatility (greater impact) in comparison to
positive return shocks of the same magnitude (Ederington and Guan 2010).
Generally, investors “within a given time period” require a greater expected return from
riskier securities (Glosten et al. 1993) in order to be compensated for taking on the
additional risk. In other words, a volatility shock could significantly increase risk
premium (Grossman 1988). As a result, during a stock market crash, which is an
example of a negative return shock, asymmetric volatility is most apparent (Wu 2001).
Wu (2001) states that during a stock market crash, “a large decline in stock price is
associated with a significant increase in market volatility”. Nevertheless, “whether or not
investors require a larger risk premium on average for investing in a security during
times when the security is more risky remains an open question” (Glosten et al. 1993).
Accordingly, the risk - return relationship across time is controversial and yields
conflicting results. Leverage hypothesis (Black 1976; Christie 1982; Glosten et al. 1993)
and volatility feedback theory (Pindyck 1984; French et al. 1987; Schwert 1989;
Campbell and Hentschel 1992) propose a positive relationship between expected return
and conditional volatility during a shock. “Although, to many, “leverage effects” have
become synonymous with asymmetric volatility, the asymmetric nature of the volatility
response to return shocks could simply reflect the existence of time varying risk
premiums” (Pindyck (1984) and French et al. (1987) cited in (Bekaert and Wu 2000).
According to the leverage hypothesis: substantial fall in the stock price (negative stock
return) means a fall in the equity value and increased financial leverage at the average
firm level making the stock riskier and increasing the equity’s volatility. Conversely, 14 Negative relationship is showed by continuous time stochastic volatility models of Bakshi et al. (1997) and Bates (1997).
31
during a positive return shock, leverage reduces resulting in reduced volatility. The
volatility feedback hypothesis states that if volatility is priced, a return shock causing an
expectation of future volatility to rise will also increase the required rate of return or
expected return (to compensate the investor) and consequently result in a fall in current
stock prices (negative return). Such volatility feedback “partially offsets a positive return
and augments a negative return” (Ederington and Guan 2010).
Accordingly, during a risky period, both leverage and volatility feedback theories reflect
that rational risk averse investors require larger risk premiums (increased expected
return) to be compensated for added risk. Therefore, during stock market crashes,
increased expected return will cause sharp decline in the current stock returns which will
adversely affect the abnormal return15. Studies which have empirically found a positive
relationship between expected excess return (risk premium) and conditional variance
include French et al. (1987), Chou (1988) and Campbell and Hentschel (1992).
Conversely, Glosten et al. (1993) state that investors may not necessarily require higher
risk premiums during risky periods.
“A larger risk premium may not be required, however, because time periods
which are relatively more risky could coincide with time periods when investors
are better able to bear particular types of risks. Further, a larger risk premium
may not be required because investors may want to save relatively more during
periods when the future is more risky. If all the productive assets available for
transferring income to the future carry risk and no risk - free investment
opportunities are available, then the price of the risky assets may be bid up
considerably, thereby reducing risk premium” Glosten et al. (1993) .
15 Abnormal return is the difference between actual return and expected return (see chapter 3 for more explanation).
32
In support, Bekaert and Wu (2000) state that the legitimacy of time-varying risk
premium is questionable when the relationship between markets expected return and
market conditional volatility is not positive. Poterba and Summers (1986) argue that
transitory impacts on volatility changes will result in the market making insignificant
adjustments to risk premium. It follows that if the impact of a shock on volatility is
transitory, no significant changes to the risk premium will occur.
Additionally, Basher et al. (2007) empirically tested the time-varying risk return
relationship in Bangladesh stock market using Generalized Auto Regressive Conditional
Heteroskedasticity in Mean (GARCH – M) model from 1 September 1986 to 30 January
2002. The study finds “significant relationship between conditional volatility and stock
returns”. However, “risk – return parameter is found to be sensitive to choice of samples
and frequencies of data”. Generally, a negative relationship between risk and return was
found supporting the findings of Glosten et al. (1993). Basher et al. (2007) state that
although the negative risk-return relationship is inconsistent with the portfolio theory, “it
is possible theoretically in Emerging Markets (EM) as investors are better able to bear
risk at times of particular volatility (Glosten et al. 1993)”.
Furthermore, Fang (2001) examined the stock returns of Taiwan stock exchange by
employing ARCH (3) - model and using daily data for the period 1995 to 1998. The
study finds that during the Asian financial crisis (1997 - 1998), stock market volatility
increased but risk premium remained constant implying that increased volatility was not
compensated by increased risk premium and larger returns. Conversely, Celikkol et al.
(2010) investigated the impact of LB bankruptcy on the volatility structure of Turkey’s
stock market using Istanbul Stock Exchange (ISE) - 100 index by examining the period
before LB bankruptcy (04 March 2008 to 14 September 2008) and post-bankruptcy (16
September 2008 to 07 April 2009) using ARCH – GARCH models. The study finds
average returns of investors and volatility of ISE - 100 index increasing in the
bankruptcy period in comparison to the pre-bankruptcy period due to positive risk and
33
return relationship. Similarly, Choudhry (2000) find both the mean return and volatility
spillover increasing after the 1987 crash.
Other studies which have found a negative relationship between the risk and return
include Fama and Schwert (1977), Campbell (1987), Turner et al. (1989), Breen et al.
(1989), Baillie and DeGennaro (1990), Pagan and Hong (1991), Nelson (1991), Glosten
et al. (1993), Whitelaw (1994), Boudoukh et al. (1997), De Santis and Imrohoroglu
(1997) and Whitelaw (2000). However Lee et al. (2001), unlike the assumption of asset
pricing model, find no relationship between expected return and expected risk. Also,
Baillie and DeGennaro (1990), Choudhry (1996), De Santis and Imrohoroglu (1997),
Lee et al. (2001) and Shin (2005) document weak (insignificant) relationship between
risk and return based on parametric GARCH – M model.
2.4 AN OVERVIEW OF THE IMPACT OF STOCK MARKET CRASHES
Stock market crashes are described as infrequent events significantly affecting
shareholder’s wealth (Ammann and Kessler 2008) under unstable and soaring
information asymmetry market conditions. Therefore, such an event offers an exclusive
opportunity for better understanding the dynamics of stock price “and the relationship
between investors’ trading strategy, volatility, volume and returns under high
asymmetric information environment” (Visaltanachoti and Luo 2009).
Ammann and Kessler (2008) “define a crash as a substantial and rapid price decline
preceded and followed by a period of constant prices”. In support, Brisbois et al. (2000)
state that “crashes are several orders of magnitude more frequent than expected from the
distribution of daily fluctuations”. Fama and French (1989) find crash causing changes
in expectations about growth after a period of extended growth market and increasing
the expected returns which results in an increased discount rate. Accordingly, Rendu de
Lint (2002) defines crisis as an abnormal decline in excess returns.
34
Stock market crashes have lower tails implying greater negative movements while
market booms have upper tails implying greater positive movements (Cotter 2006).
Cotter (2006) investigated the extreme nature of returns during the financial crisis using
the Extreme Value Theory for a range of US, European and Asian equity market indices.
The study finds positive fat tailed indices characterized by Frechet distribution. In
support, for China’s stock market, Lee et al. (2001) also finds “a fat-tailed conditional
distribution of returns, which implies that large changes in speculative prices are
expected relatively often”.
Additionally, Cotter (2006) also finds extreme return levels associated with market
crashes as being more severe in magnitude than the booms. Asian markets experience
greatest frequency and severity of extreme return implying “largest propensity for
experiencing crashes and booms” (Cotter 2006). In support, Chen et al. (2003), Law
(2006) and Ederington and Guan (2010) find negative return shocks leading to a greater
fall in national stock return together with an increased volatility in comparison to an
equal magnitude of positive return shocks.
Furthermore, Kryzanowski et al. (1995) examined the impact of the Canadian stock
market crash of 1987 on the performance of screen – sorted portfolios abnormal returns,
volatility and residual risk premium. The study employs an event study approach and
uses GARCH – M model to check the robustness of the results. The study finds that
there is an inverse relationship between the behaviors of beta sorted portfolios and
systematic risk over different time intervals. Also, after the crash, the lowest beta sorted
portfolio had a statistically significant increase in its residual risk premium and higher
negative abnormal returns.
Moreover, Wang et al. (2009) investigated the effects of various stock, firm and industry
characteristics in influencing individual stock returns during eight major stock market
35
crashes from 31 December 1962 to 31 December 2007. Event study methodology and
multivariate regression were used to study a large sample of US firms excluding Utilities
and Financial firms. The study finds that during stock market crashes more value is lost
for those stocks associated with higher betas, larger capitalization, lower levels of
illiquidity and more return volatility a year prior to the event date.
Wang et al. (2009) also find that in most stock market crashes, there is “a positive
momentum effect for the cumulative stock returns earned one-week prior to the crash
date and a negative reversal effect for the cumulative stock returns earned three months
and three years prior to the crash date”. In other words, stocks having higher returns a
week prior to the crash date also had higher returns on the crash day while stocks having
higher returns three months and three years prior to the crash date suffered more losses.
Also, high tech firms were found to lose more value in comparison to other firms in
other industries in most crashes.
Moreover, “market places are highly correlated around stock market crashes” (Brisbois
et al. 2000). Bartram and Bodnar (2009) also find significantly increased cross-country
correlation during the peak crisis period in comparison to the post crisis period of GFC.
Therefore, “there is a growing consensus that analyzing financial contagion16 is essential
in understanding financial crisis, especially in view of the increasing degree of
integration of international financial markets” (Caporale et al. 2006). Accordingly, a
negative return shock will be transmitted to other stock markets due to contagion effects.
Choudhry (2000), Chen et al. (2003), Hsin (2004), Gklezakou and Mylonakis (2009),
Cheung et al. (2010) and Kenourgios et al. (2011) find contagion effects from the crisis
country to other countries. Conversely, Ahlgren and Antell (2010) examine share prices
from developed as well as emerging economies and find evidence of short-term linkages
in times of crisis but not contagion. In the next section, contagion effects from the Asian
financial crisis and the GFC (including LB Bankruptcy) is discussed respectively. 16 A financial contagion refers to the immediate significant spread of shocks from a particular country's financial market to other countries financial markets during a financial crisis (Edwards et al. 2003).
36
2.4.1 CONTAGION EFFECTS FROM THE ASIAN FINANCIAL CRISIS
2.4.1.1 THE IMPACT OF THE ASIAN FINANCIAL CRISIS ON STOCK MARKET
PERFORMANCE
The Asian financial crisis started in 1997 from the collapse of Thai Baht and proliferated
to other economies. It triggered instability in the “stock and foreign exchange markets in
Asia since July 1997” (Fang 2001) and had a profound impact on financial markets
globally (Visaltanachoti and Luo 2009). As a result of financial contagion, the collapse
of Thai baht during the currency crisis in Thailand resulted in financial instability
spreading quickly throughout East Asia and then to Russia and Brazil and developed
markets in North America and Europe (King 2001).
Following the Asian Crisis of 1997, several papers have investigated its impact on stock
market volatility (Fang 2001; Bautista 2005; Cuñado et al. 2006; Law 2006;
Hammoudeh and Li 2008; Nam et al. 2008; James and Karoglou 2010) and the
subsequent contagion effects. The research findings generally show that the Asian crisis
substantially increased volatility in the stock markets. Bautista (2005) finds high
volatility episodes (for East Asian stocks) with differing country to country magnitude
for Hong Kong, Korea, Indonesia, Singapore, Malaysia, Philippines and Thailand.
James and Karoglou (2010) find stock market volatility in Indonesia significantly
increasing during the Asian crisis. Cuñado et al. (2006) find persistence of volatility for
the stock markets in Korea and Thailand during the Asian flu. Law (2006) find
prolonged stock market volatility and diminishing investor confidence in Malaysia (the
Kuala Lumpur Stock Exchange). These were reflected by a huge fall in the stock index.
Although the stock return volatility was found to be falling by March 2003, it had not
reverted to the level prior to the crisis (Law 2006).
Hammoudeh and Li (2008) find the Asian crisis along with other global events (1998
Russian crisis, the collapse of the oil prices at the end of 1998, the 2000 adoption of new
37
oil pricing mechanism by Organization of Petroleum Exporting Countries and
September 11th attack) causing sudden changes in volatility in the Gulf Cooperation
Council markets.
2.4.1.2 CONTAGION EFFECTS
Baig and Goldfajn (1999) tested for contagion effects during the Asian Financial crisis.
They found evidence in favor of substantial contagion in the foreign debt markets
between the financial markets of Thailand, Malaysia, Indonesia, Korea and the
Philippines whereas the evidence on share market contagion was more tentative. Large
and significant cross country correlations in equity markets was also found. Tai (2007)
examined the Asian emerging share markets (India, Korea, Malaysia, Philippines,
Taiwan and Thailand) for contagion effects and found strong positive relationship
between stock returns originating from the domestic stock market and its foreign
exchange market during the crisis.
Nam et al. (2008) state that “the stock prices of Asian EM have been at tandem with
sharp moves of the US market since the 1997 financial crisis”. Consequently, Nam et al.
(2008) investigate the price and volatility spillover effects, which is based on new
information on stock prices originating from the US markets, being transmitted to five
Pacific Basin EM (Hong Kong, Singapore, South Korea, Malaysia and Taiwan). Nam et
al. (2008) used Exponential GARCH model to compare spillover effects before and after
the crisis period. After the crisis, unlike the Malaysian market, Hong Kong, Singapore,
South Korea and Taiwan experienced a marginally stronger price spillover effect. On the
other hand, after the crisis, the influence of US shocks on market volatility transmission
became less pronounced in all markets except for the Korean market. Korea and
Malaysia implemented different approaches in addressing the crisis which resulted in
post crisis opposite shifts in price and volatility spillovers.
38
2.4.2 CONTAGION EFFECTS FROM THE GFC AND LB BANKRUPTCY
2.4.2.1 THE IMPACT OF THE GFC AND LB BANKRUPTCY ON STOCK
MARKET PERFORMANCE
The US 2007–2009 GFC started from the collapse of the sub-prime mortgage market
and the subsequent failure of the banking markets (Syllignakis and Kouretas 2011). The
stock market decline from the GFC is the second largest decline since the Great
Depression of 1929. “The crisis has driven down equity levels across the globe in nearly
every country, sector and industry” (Bartram and Bodnar 2009). According to Bartram
and Bodnar (2009), since all equity levels were adversely affected, equity investors
could not benefit from financial advices on diversifications.
Furthermore, despite the ongoing mortgage and banking crisis since early 2007, the
equity market reaction until July/August 2008 was basically second order (Bartram and
Bodnar 2009). The “real equity market action (collapse)” begins with the bankruptcy of
LB on 15 September 2008 (Bartram and Bodnar 2009). Bartram and Bodnar (2009) find
the period from 15 September 2008 following LB bankruptcy through the end of
October 2008 as marking the “heat of the crisis” with index levels falling dramatically
and price volatility skyrocketing universally. The study finds that most indices
experienced substantial declines for the year during the 31 trading days from 12
September 2008 to 27 October 2008 and this was identified as the crisis period (Bartram
and Bodnar 2009).
In support, Samarakoon (2011) identifies the most chaotic period of melt down during
GFC as a 6 month period from September 2008 to early March 2009. Samarakoon
(2011) states that during the chaotic period, there was a decline in the US Stock market,
EM and frontier markets by 43%, 50% and 60% respectively. According to Eichler et al.
(2011), a further deterioration of the long-term perspectives, which added to the long-
term crisis risk (increasing the risk index to 100%), particularly resulted from the events
in September 2008. Eichler et al. (2011) states that:
39
“In early March 2008, when Bear Stearns & Co. was close to bankruptcy, the
risk index is above 85%. The situation worsens till July, when the perceived risk
is well-above 90%. This clearly indicates the upcoming defaults of the included
banks that occurred in September 2008 when the default risk materializes and
the risk index is almost 100%”.
Accordingly, LB bankruptcy resulting from the mortgage crisis caused “deep fall on
stock market price” and had the most significant impact on the stock market (Chong
2011). The financial sector was identified as the epicenter of the GFC and the collapse
of LB marks the anticlimax era in the stock market performance around the world
(Bartram and Bodnar 2009).
2.4.2.2 CONTAGION EFFECTS
All the equity markets (US, Developed markets excluding North America and EM)
suffered sharp declines with immense wealth losses during the financial crisis of 2008 –
2009 and the US market had a “higher volatility measure in all periods” from 31
December 2006 to 27 February 2009 (Bartram and Bodnar 2009). Chong (2011)
investigated the impact of the subprime crisis on the US S&P 100 return and volatility
using daily data from May 2006 to December 2009. The study finds that LB bankruptcy
after the subprime crisis generally had a significant effect on the US stock market
volatility but not on the stock returns. Also, LB bankruptcy shock was found to have a
transitory impact on the stock market volatility with very slow rate decay due to the
persistence measure (α +β ) being very close to unity.
Bartram and Bodnar (2009), find a significant increase in average cross-country
correlation changes at the market index level for the Developed Markets [US, United
Kingdom (UK), Europe ex UK, Japan and Pacific ex Japan] and the EM [three EM
regions of EM Asia, EM Latin America and EM Middle East Africa (MEA)] during the
40
peak crisis period (12 September 2008 to 27 October 2008) of GFC. Syllignakis and
Kouretas (2011) also find significant evidence in support of contagion effects during the
2007-2009 financial crises amongst “the US and German stock returns and the Central
and Eastern European (CEE)17 stock returns”.
In the crisis period, the highest correlation was found amongst the developed markets
(Bartram and Bodnar 2009). Additionally, cross market correlations between developed
and EM was also the highest (Bartram and Bodnar 2009). The findings imply that “this
crisis itself was really a problem in Developed markets, and that it is more sensitive to
linkages across Developed markets than EM” (Bartram and Bodnar 2009). Furthermore,
Angkinand et al. (2009) examined the extent of interdependence and spillover effects for
various sub periods from January 1973 to February 2009 amongst the US and 17
advanced economies 18 national stock market returns. Angkinand et al. (2009) find
increased interdependence and spillover effects on economies after the U.S financial
crisis (summer of 2007), particularly following LB bankruptcy (in September 2008).
Although the extent of interdependence between the US and 17 advanced countries
varied in the earlier decades, it became fairly homogeneous after the US financial crisis
(Angkinand et al. 2009).
Park (2010) investigated the impact of global economic crisis and its recovery for the
stock markets of sixteen countries using descriptive analysis. The findings indicated
significant changes in the stock return patterns after the crisis. Decrease in the mean
return and an increase in volatility, skewness and kurtosis were observed (Park 2010).
17 The CEE stock markets were namely Czech Republic, Estonia, Hungary, Poland, Romania, Slovakia and Slovenia (Syllignakis and Kouretas 2011). 18 The 17 advanced countries were Australia, Austria, Belgium, Canada, Denmark, France, Germany, Hong Kong, Italy, Japan, Netherlands, Norway, Singapore, Spain, Sweden, Switzerland and UK (Angkinand et al. 2009).
41
Furthermore, Nikkinen et al. (2008) find that despite globalization (suggesting
uniformity in responses to shocks) the exposure of regions to shocks varies depending
on the extent of their integration with the international markets. It states that, “the less
integrated regions (e.g. Middle East and North Africa) are with the international
economy, the less exposed they are to shocks”. The findings of Nikkinen et al. (2008)
are supported by (Dufrenot et al. 2011).
Dufrenot et al. (2011) investigated the impact of the subprime crisis on the Latin
American stock market volatility using Markov - switching model. The study finds
heterogeneous response of Latin American stock markets to the US financial market
stress (bad news on LB bankruptcy, US financial institutions equity write-downs or
housing market developments). Dufrenot et al. (2011) find that although all markets
suffered increased volatility in equity prices, the extent of volatility being transmitted to
the stock markets varied based on the degree of interdependence with the US market.
Since Mexico was more closely linked to the US financial market, it was most
susceptible to the contagion effects stemming from the US financial stress. During the
crisis, in addition to the regional volatility from regional equity markets, Mexico and
Chile suffered increased stock market volatility changes caused by the US financial
stress. On the contrary, Colombia, Peru and Brazil appeared more prone to regional
volatility in comparison to volatility from US financial stress.
Longstaff (2010) empirically investigated the pricing and contagion effects of subprime
asset- backed collateralized debt obligations (CDO) on other markets using the proxy of
ABX subprime indexes data. The findings strongly support contagion in the financial
markets transmitted primarily via “liquidity19 and risk- premium20 channels rather than
through a correlated market channel”. The results also imply that during the crisis,
19 “In this mechanism, a shock to one financial market signals economic news that is directly or indirectly relevant for security prices in other markets” (Longstaff 2010). 20 “In this mechanism, financial shocks in one market may affect the willingness of market participants to bear risk in any market. Thus, prices in all markets may be affected as equilibrium risk premia adjust in response” (Longstaff 2010).
42
significant price discovery took place in the subprime market since ABX index returns
were able to forecast changes in stock returns and Treasury and corporate bond yields to
an extent of three weeks earlier. In addition, increased participation of foreign investors
in CEE stock markets and financial liberalization were identified as attributes for high
degree of correlations during the recent stock market crash. Conversely, Syllignakis and
Kouretas (2011) find that “for the cases of the Asian and Russian crises (1997–1998)
and the dot-com bubble, the contagion effect hypothesis cannot be accepted throughout
the cross section” because “the recent crisis was expected to be more related to the CEE
markets than the earlier Asian crisis”. The main findings of the paper imply that the
impacts varied due to differing origins and type of the crises (Syllignakis and Kouretas
2011).
2.4.3 CONTAGION EFFECTS, GLOBAL AND LOCAL RISK FACTORS
As a result of international financial market linkages and potential contagion effects
investors incorporate global as well as local risk factors in anticipating prospective
returns and in forming investment decisions. Rendu de Lint (2002) examined the
changes in the risk profiles 21 of Mexico, Indonesia, Korea, Malaysia, Philippines,
Singapore, Taiwan and Thailand stock markets over time and during Asian crisis. They
find that in periods approaching a crisis, apart from global risk factors, investors also
incorporate local risk factors in anticipating future excess returns.
On the other hand, Choudhry (2000) finds that prior to the 1987 crash; only domestic
news influenced domestic market mean returns and volatility. However, they find that
post-crash, foreign markets news significantly influenced the domestic markets. In
addition, Cuñado et al. (2006) state that “global events impact all countries, but this
impact is generally short lived and does not cause structural changes in the economies.
Changes in the structure and level of volatility/instability come mainly from local
events, which in most cases are indeed associated with financial liberalization
processes”. 21 Refers to threats to which the stock market is exposed to.
43
2.5 CONCLUSION
Financial liberalization has resulted in stock market interdependence. However, there are
mixed results on the impact of financial liberalization on stock market volatility. In the
world equity market, the US stock market plays the most prominent role in influencing
other markets return and volatility. Volatility asymmetry is found in the US stock
market particularly during stock market crashes. However, the risk - return relationship
across time is controversial and yields conflicting results. Industry specific turmoil
particularly arising from the financial sector is perceived to be more dangerous and
having a larger impact than country specific crises. The US financial sector was the
epicenter of the recent GFC and the collapse of LB marks the anticlimax era in the stock
market performance globally. Following LB bankruptcy, the correlation within
industries and across US industries as well as regions increased (Bartram and Bodnar
2009). This implies that LB bankruptcy also influenced other sectors apart from
financials.
Motivated by literature, this study extends the work of Pichardo and Bacon (2009),
Raddatz (2010) and Dumontaux and Pop (2009) by thoroughly investigating the impact
of LB bankruptcy on all NYSE sectors and financial industries. This study examines
whether or not the financial sector and the diversified financial industry were the most
significantly adversely affected from LB bankruptcy. The next chapter provides
discussions on the data, methodology and hypotheses development.
44
CHAPTER 3: DATA, METHODOLOGY AND HYPOTHESES DEVELOPMENT
3.0 INTRODUCTION
This study uses an event study methodology to examine the US stock market
performance at sector and financial industry level during the event of LB bankruptcy.
Other studies that have used event study methodology to examine single day events
include Kryzanowski et al. (1995), Pichardo and Bacon (2009), Raddatz (2010)
Dumontaux and Pop (2009) and Mio and Fasan (2012). Kryzanowski et al. (1995) used
it to examine the impact of the Canadian stock market crash of 1987 on the performance
of screen-sorted portfolios’ abnormal returns, volatility and residual risk premium, while
Pichardo and Bacon (2009), Raddatz (2010) and Dumontaux and Pop (2009) used it to
examine the impact of the LB bankruptcy on the performance of financial firms. Mio
and Fasan (2012) used an event study methodology to examine whether corporate social
performance had any impact on corporate financial performance due to the LB
bankruptcy. The authors analyzed the S&P 500 market index’s constituent non-financial
companies’ stock prices prior to and during the LB bankruptcy announcement.
The event date in this study is defined as Monday, 15 September 2008 22 . LB
bankruptcy23 was announced on 15 September 2008 (Dumontaux and Pop 2009). The
choice of the event date is motivated by the following reasons. (1) “September 15, 2008
has been proclaimed Wall Street’s worst day in seven years. The Dow Jones Industrial
average lost more than 500 points, more than 4%, which is the steepest fall since the day
after the September 11th attacks” (Pichardo and Bacon 2009). (2) On the date of LB
bankruptcy announcement, which is the event date, LB had total assets of $639 billion
making it the largest failure in the US history. This stimulates the need to examine the
stock market reaction to the catastrophic news on the immense failure of the largest
fixed interest security dealer, LB. (3) Despite the ongoing mortgage and the banking 22 Other studies that have used 15 September 2008 as the event date for examining the performance of stocks during LB bankruptcy include Pichardo and Bacon (2009) and Dumontaux and Pop (2009). 23 The bankruptcy petition no. was #08-13555 (Dumontaux and Pop 2009).
45
crisis since early 2007, the equity market reaction until July/August 2008 was basically
second order. The “real equity market action (collapse)” begins with the bankruptcy of
LB on 15 September 2008 (Bartram and Bodnar 2009). “Despite the overall volatility in
international liquidity markets, the 15 September 2008 episode stands out because of its
magnitude and fast onset” (Raddatz 2010). Therefore, the LB bankruptcy announcement
on 15 September 2008 triggered the GFC that makes it an event worth being
investigated. (4) Pichardo and Bacon (2009), Raddatz (2010) and Dumontaux and Pop
(2009) have also used 15 September 2008 as the event date to examine the impact of LB
bankruptcy on the performance of financial firms. Therefore, this study’s findings will
be comparable to the mentioned studies and contributions to the extant literature could
be made in a more meaningful way with the use of a large sample (this study undertakes
a detailed analyses of both financial and non-financial firms at sector and financial
industry level) and robust event study approaches (see next section for a detailed
discussion of these approaches).
In this chapter, discussion begins with a detailed description of the dataset used in this
thesis, the data source, the sample selection procedures and the justifications for the data
used (see Section 3.1). Event study approach is used to evaluate the performance of the
sectors and financial industries. In Section 3.2, the event study approach, which is based
on the standard OLS MM and a variant of the standard OLS MM, is discussed. This is
followed by Section 3.3 that discusses the robustness test conducted to validate the event
study results. Section 3.4 develops the research hypotheses. It provides a discussion on
LB bankruptcy, the financial sector, the diversified financial industry and accordingly
states the two hypotheses. Section 3.5 concludes this chapter.
3.1 DATA AND SAMPLE SELECTION
This study investigates the impact of LB bankruptcy announcement (event date) on the
performance of NYSE sectors and financial industries. Event study approach is used to
examine the performance of financial industries and sectors during LB bankruptcy. The
46
event study approach is based on the theory of market efficiency. As a result, this study
examines share price returns24 data for 481 firms (488 firms under the standard OLS
MM) grouped into sectors and financial industries during the event window. The event
window chosen for this study is from 9 September 2008 to 15 September 2008 (refer to
page’s 62 to 63 for justification on the choice of this event window). The S&P 500 index
is used as a proxy for the US stock market. The 9 March 2011 S&P 500 composition
firm data list25, which contained information on company name, ticker symbol and
Global Industry Classification Standards sector classification (GICS), was downloaded
from Standard and Poor’s website. Next, each of the company’s ticker symbols was used
to download their respective closing share price from yahoo finance. Following this,
each of the firm’s data was distributed into sectors and financial industries according to
their corresponding GICS.
Based on the above categorization, CD, CS, Consumer Services, Eng, Fin, HC, Ind, IT,
Mat, Uti and TS sectors were formed. The Consumer Services sector, which had only
one firm, was dropped from the sample. Therefore, 10 sectors are analyzed in this study.
The four financial industries analyzed in this study are namely banks, diversified
financial, insurance and REIT. Table 3.1, Panel A provides the sample selection
procedures under the standard OLS MM and the variant of the standard OLS MM. Any
firm with missing data during the estimation period or the event period was removed
from the sample to avoid biasness in the results. As a result, the sample used in the
standard OLS MM consists of a total of 488 firms; 78 firms in the CD sector, 37 in CS,
39 in Eng, 81 in Fins, 50 in HC, 62 in Inds, 70 in IT, 30 in Mats, 33 in Uti and 8 in TS
sector. On the other hand, the sample used in the variant of the standard OLS MM
consists of a total of 481 firms; 77 firms in the CD sector, 37 in CS, 38 in Eng, 79 in
Fins, 49 in HC, 62 in Inds, 69 in IT, 30 in Mats, 33 in Uti and 7 in TS sector. Table 3.1,
Panel B provides the sample size for each of the financial industries used in the variant
24 Equation 2 on page 54 provides the formula to compute share price returns. 25 This list does not include the ticker for LB and LB was not categorized as either a financial sector or a Diversified financial industry. Therefore, LB returns is not included in the returns of the financial sector or Diversified financial industry.
47
of the standard OLS MM. There were 20 firms in banking industry, 21 firms in
diversified financial industry, 20 firms in insurance and 16 firms in REIT industry.
The data selected for this thesis has the following advantages. Firstly, LB was listed on
the NYSE and traded as part of the S&P 500 index prior to its bankruptcy. This
motivated the choice of using the S&P 500 index data to analyze the impact of LB
bankruptcy at the sector and financial industry level. Also, it is appropriate to conduct an
event study in the US stock market as Pichardo and Bacon (2009) find US market semi
strong form informational efficient during LB bankruptcy. Therefore, NYSE security
returns are anticipated to be instantaneously reflective of the effects of LB bankruptcy
on the firm’s value. Secondly, this study uses daily returns to conduct an event study.
Brown and Warner (1985) find the use of daily returns more powerful than monthly
returns. Thirdly, the choice of the event window and its estimation period for the
standard MM, which is employed in this study, has been adopted from Pichardo and
Bacon (2009) to extend its work and to add to the extant literature. This will make the
findings of this study directly comparable to those of Pichardo and Bacon (2009).
48
Table 3.1: Sample Data Set
Note: This table provides the sample data used in this thesis. Panel A details the sample selection procedure for each of the 10 sectors under the standard OLS MM and the variant of the standard OLS MM respectively. Panel B provides the sample size for the financial industries used in the variant of the standard OLS MM.
Panel A: Sample selection procedures for each of the 10 sectors
Standard OLS MM Variant of the standard OLS MM
Sectors Total Firms
Number of Firms
Removed
Ticker Symbol of
Firms Removed
Firms Remaining
Number of Firms
Removed
Ticker Symbol of
Firms Removed
Firms Remaining
CD 79 1 SNI 78 2 SNI, TWC 77
CS 41 4 DPS, LO, MJM, PM 37 4 DPS, LO,
MJM, PM 37
Eng 39 - - 39 1 SE 38
Fin 82 1 BRKb 81 3 BRKb, CMF, DFS 7926
HC 51 1 CFN 50 2 CFN, COV 49
Inds 62 - - 62 - - 62
IT 72 2 MMI, V 70 3 MMI, V, TDC 69
Mats 30 - - 30 - - 30
Uti 35 2 PEG, QEP 33 2 PEG, QEP 33
TS 8 - - 8 1 PCS 7
Total 499* 11 488 18 481
Panel B: Sample size for the financial industries
Industries Sample Size
Banks
20
Diversified Financials
21
Insurance 20
REIT
16
Total 77
26 Amongst the 79 financial firms, based on GICS, only 77 firms could be classified into the four respective financial industries.
49
3.2 EVENT STUDY
This study employs an event study approach to examine the performance of the US
disaggregated sectors and financial industries during LB bankruptcy27. In this section,
the rationale behind the usage of the event study approach is discussed. This is followed
by discussions on methods to conduct the event study and its common applications.
Next, various approaches to compute normal period return is discussed along with their
respective advantages and disadvantages. Additionally, explanations/reasons on the
rigorous and widespread usage of the market model (MM) are provided. Furthermore,
the rigorousness of the MM approach over other statistical, non-regression based models
and economic models is discussed. Finally, the approaches employed for estimating
normal period returns in this study is provided along with the justifications for their
usage.
The event study approach is a widely used analytical technique in the field of finance to
examine the impact of significant event(s) on the performance of financial stock prices.
Both firm specific and economy wide events are investigated using this approach.
According to Schweitzer (1989) there are two types of events which are examined using
event study approach namely a single event with one time occurrence and events with
frequent occurrence such as earnings announcements. Based on an event with single
occurrence, a number of studies examined the impact of the financial crisis on the
performance of stock returns (Kryzanowski et al. 1995; Pichardo and Bacon 2009;
Raddatz 2010; Dumontaux and Pop 2009; Mio and Fasan 2012).
Based on the theory of market efficiency, event studies reflect the immediate effect of a
particular event on the performance of stock returns. “Event studies provide a direct test
of market efficiency” (Brown and Warner 1980). The theory of market efficiency
suggests that security prices and their returns are anticipated to be instantaneously
27 Kryzanowski et al. (1995) employs an event study approach to examine the impact of the Canadian stock market crash of 1987 on the performance of screen – sorted portfolios.
50
reflective of the effects of firm specific and economic wide events (MacKinlay 1997).
The event study approach is employed to examine the reaction of security prices
surrounding the event under investigation. To ascertain the impact of event(s), expected
normal period returns (referred as normal returns), which would have eventuated in the
absence of the event, are compared with the ex-post actual realized returns that
incorporates the information of the event. The difference between the two is known as
abnormal returns and is attributed to the event under investigation. In this manner, the
event study approach is considered an effective way of analyzing the impact of firm
specific or economic events on the firm’s value immediately during the event period.
Assuming semi-strong form of efficiency in the market, MacKinlay (1997) finds
evidence strongly supporting “the hypothesis that earnings announcements do indeed
convey information useful for the valuation of firms”. Therefore, “the use of event study
methodology has become a commonplace with respect to the detection of wealth effect”
(Brown and Warner 1985) and “dominate empirical research” in corporate finance
(MacKinlay 1997).
MacKinlay (1997) states that for a given security (firm) i and event date t, the abnormal
return is calculated as:
(1)
Where is the abnormal return for security i at time t, is the actual realized return
for security i at time t and is the normal return for security i at time t
conditional upon the information for the normal return model which will be discussed
next.
Abnormal returns could be either positive or negative. A positive abnormal return for
any particular security i at time t during the event period implies that an event had a
positive impact on the performance of the security’s return. Conversely, a negative
51
abnormal return for any particular security i at time t during the event period implies that
an event adversely affected the security’s return performance.
Normal period returns can be computed using various approaches grouped into statistical
and economic categories. The statistical approach to compute normal period returns
dominates event studies (MacKinlay 1997). Under the statistical approach, there are
three ways in which the daily abnormal return for each security could be calculated.
These are mean adjusted returns, market adjusted returns and Ordinary Least Squares
(OLS) MM. Both mean adjusted returns and OLS MM techniques use information from
the estimation period to compute expected normal period returns. An estimation period
is any time period apart from the “period immediately surrounding the event date”
(Peterson 1989). The normal period return using mean adjusted returns is the simple
average of security i’s daily return during the estimation period. The OLS MM uses the
regression results of from the estimation period to predict normal period
returns. are the regression output obtained by regressing each security’s price
return on the concurrent stock market portfolio index. S&P 500 index, the Centre for
Research in Security Prices (CRSP) Value Weighted Index and the CRSP Equal
Weighted Index are mostly used as a proxy for return on the market portfolio index
( ). Refer to Equation 3 of this thesis (page 56) on OLS MM to see how the expected
return is calculated. Unlike the mean adjusted returns and OLS MM techniques, the
market adjusted return approach to event study does not have any estimation period.
“Market adjusted return model can be viewed as a restricted market model” by
constraining to zero and to one (MacKinlay 1997). Due to the pre specification of
the and parameters, there is no need for the estimation period (MacKinlay 1997).
Under the market adjusted return model, the best predictor of the normal period return is
said to be the current return on the market portfolio index. Therefore, the current event
period market return is used as a proxy for the expected normal period return.
52
Amongst the event study approaches, the MM is the most widely used model. Cable and
Holland (1999) provide support for regression based approach to event study and
particularly the use of MM over non-regression based models. Cable and Holland (1999)
state that now linear MM is frequently used for regression - based studies. In earlier
research, Cable and Holland (1996) presented a model - selection framework and
showed preliminary preference for the MM in comparison to the non-regression models.
Cable and Holland (1999) find that in confirmation to results from Brown and Warner
(1985), the “best practice would seem to mandate the use of regression based models
and in particular the market model”. The benefit of using MM over the two commonly
used economic models; CAPM and Arbitrage Pricing Theory (APT) are as follows.
MacKinlay (1997) states that the use of MMs can eliminate the sensitivity in the results
arising from CAPM restrictions. On the other hand, the main potential advantage of
using APT motivated model is to eliminate the biasness from the use of CAPM.
However, the statistically motivated models also eliminate this biasness. Also, minimal
gains are achieved with the use of APT motivated model over MM. Therefore, statistical
methods to event study and particularly the MM is prevalent for event studies
(MacKinlay 1997).
The MM is considered a more robust approach to event study not only in comparison to
non – regressive (Cable and Holland 1996; Cable and Holland 1999) and economic
approaches (MacKinlay 1997) but also in comparison to the mean adjusted model
(MacKinlay 1997). According to MacKinlay (1997), the MM is a better model in
comparison to the mean adjusted return model because it reduces the variation in
abnormal return “by removing the proportion of the return that is related to the variation
in the market’s return”. Brown and Warner (1985) find that the simple mean adjusted
model is outperformed by both OLS MM and market adjusted return model. Since the
market adjusted return model is a restrictive MM, MacKinlay (1997) states that to avoid
possible biasness resulting from such restrictions, it is generally recommended to use
market adjusted return models only where necessary.
53
Moreover, Brown and Warner (1985) find that the event studies conducted using daily
returns are more powerful in comparison to monthly returns. Accordingly, in this study
the investigation is conducted using daily returns. The daily stock price returns28 for
each security at time t is computed as follows:
(2)
Where is the price return on security i at time t, is the closing share price29 of
security i at time t and is the closing share price of security i on the day prior to
time t.
Additionally, for event studies conducted using daily data, prior studies generally
advocate the use of OLS MM with standard parametric tests as well specified under
numerous conditions (Brown and Warner 1985; Peterson 1989). MacKinlay (1997)
states that generally OLS model is an efficient and “consistent estimation procedure for
the MM parameters”. Therefore, this study employs the OLS MM (Brown and Warner
1985; Peterson 1989; MacKinlay 1997; Lee and Connolly 2010) introduced by Fama et
al. (1969) to investigate the effects of LB bankruptcy on each of the 10 sectors and 4
financial industries in the S&P 500 composition. Under this approach, firstly, the event
study parameters (event window and estimation period) are established. The purpose of
the estimation period is to forecast normal period returns that would have eventuated in
the absence of the event under investigation. Therefore, the event period and the
estimation window need to be carefully chosen to avoid misspecifications (of alpha and
beta parameters used to forecast normal period returns) arising from data contamination 28 Analysis is also conducted using log returns and the results are consistent to those reported in this study. Following prior literature which suggests that log CAR returns can lead to biasness in event studies, the results reported in this study are obtained using the return formula in equation 2 (Barber and Lyon 1997; Kothari and Warner 1997; Dissanaike and Le Fur 2003). 29 As a robustness check analysis was also conducted using closing price returns adjusted for dividends and stock splits. However, the results are not significantly different to those reported in this study. This is consistent with Brooks (2014) that states that for very short holding periods, ignoring dividends will have a negligent effect whereas it will have a significant impact on cumulative returns over investment horizons of several years. Since this study focuses on a short horizon, the effect is insignificant.
54
(due to the overlapping of estimation period and event window). The estimation period
needs to be distinct from the event period in order to accurately predict the normal
period returns (during the event period) that would have eventuated in the absence of the
event.
Initially, this study employs the typical event study approach based on the OLS MM
with estimation windows immediately before the event period to predict normal period
returns (this is referred as the standard MM). The investigation is carried out to
determine whether the LB bankruptcy most significantly adversely affected the
performance of the financial sector and the diversified financial industry.
Peterson (1989) states that for daily studies, the typical lengths of estimation period and
event period range from 100 to 300 days and 21 to 121 days respectively. In this thesis,
150 trading day’s estimation period (27 December 2007 to 31 July 2008) and 61 days
event window (1 August 2008 to 27 October 2008) are used. In this study, the estimation
period and event window, which are used for the standard MM, is adopted from
Pichardo and Bacon (2009). Pichardo and Bacon (2009) used these estimation
parameters to examine the impact of LB bankruptcy on the performance of 15
investment firms. Therefore, this study extends the work of Pichardo and Bacon (2009)
by examining the performance of all 4 financial industries and 10 sectors during LB
bankruptcy. The advantages of adopting the estimation parameters from Pichardo and
Bacon (2009) are as follows: (1) This study’s findings will be directly comparable to
Pichardo and Bacon (2009). (2) With the use of a larger sample size, generalization30
could be made on the pervasiveness of the impact from LB bankruptcy. The timeline on
the next page shows the 150 days estimation period and 61 days event window
employed in this study under the standard OLS MM.
30 The findings of Pichardo and Bacon (2009) were not generalizable because a smaller sample sizes of only 15 investment firms were analyzed.
55
Estimation period Event Period
Days −180 −31−30 0 +30
Furthermore, the S&P 500 index is used as the market index ( ). In the estimation
period, each of the firm’s share price returns at time t is regressed on the concurrent S&P
500 index returns. The estimated alpha and beta ( coefficients from the
regression are used to compute the expected normal period return for each
security i at time t during the event period of [−30, +30] trading days. On day “0”, which
represents the event date, LB became bankrupt. The estimated returns are considered to
be “normal returns” which are unaffected by the event.
(3)
Where is the expected normal period return for security i at time t, is the
estimated alpha for security i, is the estimated beta for security i and is the S&P
500 index return at time t.
Next, the actual realized return over the event is compared with the predicted
normal return from the above model and the difference is known as abnormal
return . Abnormal returns could be either positive or negative and it could be
either significantly affected or insignificantly affected from the LB bankruptcy.
(4)
Where is the abnormal return for security i at time t, is the actual realized return
for security i at time t and is the expected normal period return for security i at
time t.
56
The average abnormal return (AAR) for each of the sectors and financial industries at
any particular day is computed as follows:
(5)
Where is number of firms in the sector/ financial industry, is the abnormal return
for security at time and is the average abnormal return for the sector/
financial industry.
The then is accumulated through time to generate the cumulative average
abnormal return (CAAR). The CAAR is divided into the following multi day periods
[−5, 0], [−4, 0], [−3, 0], [−2, 0], [−1, 0], [0, +1], [0, +2], [0, +3], [0, +4], [0, +5], [−1, +1],
[−2, +2], [−3, +3], [−4, +4], [−5, +5]. Using these multi day periods, inferences could
be made on how the sectors and financial industries performed during a combination of
few prior and post days surrounding the event date. The CAAR for each sector at any
particular day is computed as follows:
(6)
Where denotes cumulative average abnormal return for the sector/ financial
industry over a series of time during the event window of [−4, 0] trading days, is the
first period in which the are accumulated and represents the last period in
which the are accumulated.
Furthermore, the statistical significance of abnormal returns is tested to ascertain
whether LB bankruptcy had a significant positive, negative or no impact on the
performance of sectors and financial industries. In normal circumstances (without any
event), the abnormal return will be zero. Therefore, the null hypothesis is that the mean
for the abnormal returns would be zero implying that LB bankruptcy did not
significantly affect the performance of sectors and financial industries. On the other
57
hand, the alternative hypothesis is that the mean for the abnormal returns is statistically
different from zero (positive or negative) implying that LB bankruptcy did have a
significant (positive or negative) impact on the performance of the sectors and financial
industries. This study employs the standardized cross – sectional test proposed by
Boehmer et al. (1991) to examine the statistical significance of the abnormal returns.
The Boehmer et al. (1991) standardized cross – sectional test is a hybrid test formed by
combining the two approaches of standardized residuals (Patell 1976) and ordinary cross
– sectional method (Brown and Warner 1980).
The choice of the standardized cross – sectional test approach for significance testing is
motivated by the following reasons. Firstly, Boehmer et al. (1991) find that during the
existence of event induced variances31, the standardized cross sectional test performs
better than other commonly used approaches (ordinary cross sectional test, standardized
residuals test, sign tests and method of moment’s approach) of significance testing. This
is primarily because under the standardized cross sectional test, abnormal returns are
firstly normalized by the forecasted error during the estimation period to mirror
statistical error in generating expected returns (Peterson 1989). Conversely, the ordinary
cross – sectional method (one of the commonly used approaches of significance testing)
does not take into account the variance estimates from the estimation period. This can
result in misspecification of the results as even minor increases in variances caused by
an event could cause too many false rejection of the null hypothesis of no abnormal
return (Boehmer et al. 1991). However, Boehmer et al. (1991) find that via standardized
cross sectional test approach, dividing the abnormal returns by the estimated standard
deviation that has been adjusted for forecasted error as shown in Equation 7 of this thesis
(page 60) eradicates the misspecification problem of the ordinary cross – sectional
method and “produces appropriate rejection rates when the null is true and equally
powerful tests when it is false” (Boehmer et al. 1991). Accordingly, this study employs
31There is evidence that the variance of stock returns increases for the days immediately around events such as earnings announcements (Beaver 1968; Patell and Wolfson 1979). Brown and Warner (1985) illustrate how variance increase can cause misspecification of hypothesis tests conducted using standard event study procedures.
58
the standardized cross sectional test to ensure that the rejection of the null hypothesis is
accurate.
Secondly, the sample of securities used in this study is analyzed at sector and financial
industry level using the one factor MM. Therefore, there are chances for the sample used
in this study to suffer from event date clustering32. If abnormal returns across securities
are positively correlated then such clustering will increase variance of the performance
measures (e.g. residual measures) and reduce the power of the tests. Although the event
under investigation may not influence the abnormal return, variance increases induced
by other events (e.g. mandated accounting procedures) can result in too many false
rejections of the null hypothesis33 (Brown and Warner 1980). Brown and Warner (1985)
state that if securities are randomly selected in a sample, the one factor MM provides
sufficient adjustment for dependence. However, “if instead the securities came from the
same industry group (like in the case of this study), with clustering there could be a
higher degree of cross sectional dependence in market model excess returns (abnormal
returns) and measurable misspecification …” (Brown and Warner 1985). Therefore, for
the sample used in this study, there could be a high degree of cross – sectional
dependence in the abnormal return arising from event date clustering (Brown and
Warner 1985). Nevertheless, Boehmer et al. (1991) finds that the significance test results
obtained from standardized cross sectional test (employed in this study) are robust (not
affected) in the presence of event date clustering.
In sum, according to Boehmer et al. (1991), the standardized cross sectional test
approach employed in this study is claimed to provide robust results in the presence of
32 It refers to a sample of securities suffering from other close or simultaneous spacing of events during the event period apart from the event under investigation (Brown and Warner 1980). For example, Foster (1980) and Schwert (1981) discuss government regulation or mandated accounting procedures simultaneously affecting the performance of numerous securities under examination during a particular event. 33 Due to security specific performance measures being positively correlated.
59
event induced variances and event date clustering. Under this approach, the event period
residuals are firstly standardized using the following equation.
(7)
Where is the standardized abnormal return for security i at time t, is the
abnormal return for security i on day t, is the estimated standard deviation of abnormal
returns during the estimation period for security i, T is the number of days in the
estimation period for security i, is the event period market return for day t, is the
mean market return within the estimation period and is the estimation period market
return for day j.
Since this study examines the impact of LB bankruptcy on the sample of securities at
sector and financial industry level on and around the event date, the standardized
average abnormal return (SAAR) and standardized cumulative average abnormal return
(SCAAR) is computed. This will enable the analysis to determine whether or not LB
bankruptcy had a significant impact on the performance of sectors and financial
industries on the event date. The SAAR is a cross sectional average of the sample of
securities. The SAAR will enable inferences to be made on the significance of the
impact of LB bankruptcy on a per day basis during the event window of [−30, +30]
trading days. The SAAR for each of the sectors and financial industries at any particular
day t is computed as follows:
(8)
Where N is number of firms in the sector / financial industry, is the standardized
abnormal return for security i at time t and is the standardized average abnormal
return for the sector/financial industry.
60
is then accumulated through time to generate standardized cumulative average
abnormal return (SCAAR). SCAAR is generated for multi day intervals during the event
window to make inferences on the impact of LB bankruptcy surrounding the event date.
This will provide a test of market efficiency on how the market reacted to the news on
LB bankruptcy around the event date. The SCAAR for each sector and financial industry
at any particular day t is computed as follows:
(9)
Where is the standardized cumulative average abnormal return for sectors and
financial industries over a series of time during the event window of [−30, +30] trading
days, is the standardized average abnormal return for the sector/financial
industry, is the First period in which the are accumulated and is the Last
period in which the are accumulated
Finally, ordinary cross sectional approach (Brown and Warner 1985) is employed to
compute test statistics for the SAAR and SCAAR. Accordingly, t - tests for any
particular day during the event period are computed by dividing the SAAR by its
contemporary cross – sectional standard error. The t – tests for SCAAR is computed by
dividing the mean of the SCAAR by its contemporary standard error (Boehmer et al.
1991). The calculated t - statistics (t - stats) is compared with the critical value of t at the
1%, 5% and 10% level of significance. The null hypothesis of zero abnormal return is
rejected when t-stats in absolute value > t critical value. The rejection of the null
hypothesis will imply that LB bankruptcy significantly affected the performance of
sectors and financial industries on and/or around the event date.
Moreover, during the GFC, the disadvantages of using the standard OLS MM to
examine the impact of LB bankruptcy on the performance of sectors and financial
61
industries are as follows. (1) LB bankruptcy occurs during the GFC and since there were
high levels of volatility during this period, having an estimation period immediately
prior to the event window of LB bankruptcy (which is within the high volatile period of
GFC) will result in “very imprecise estimates of excess returns and significance tests
with relatively low power” (Dumontaux and Pop 2009). How? The estimation period
return immediately prior to the event window of LB bankruptcy was subject to high
levels of volatility caused by other negative announcement events during the GFC. As a
result, the estimated beta values using this estimation period return will be biased
leading to inaccurate predictions of expected returns and abnormal returns. A
preliminary evaluation of this study’s data on financial sector returns show that the beta
during the estimation period (150 days) immediately before the event period is higher
(1.62) than the actual event period beta (1.41). This implies that the estimation period
returns immediately prior to the event period are highly contaminated by the volatility
from negative announcements during the GFC. Therefore, they are not a good estimation
parameter to predict normal period returns during LB bankruptcy34 (Dumontaux and Pop
2009). In support, Dumontaux and Pop (2009) provide evidence on the disadvantage of
using an estimation period immediately prior to the event period (−2 days to + 2 days of
LB bankruptcy) for analyses in computing abnormal returns. (2) The typical lengths of
21 to 121 days event window (Peterson 1989) would be too long as there were many
negative announcements during LB bankruptcy (not all negative news announcements
related to LB) that would impair the variance estimates of AAR. In order to control for
multiple negative news announcements during LB bankruptcy impairing the variance
estimates of AAR, a smaller event window would be appropriate (Dumontaux and Pop
2009).
Accordingly, this study attempts to address the above mentioned disadvantages of the
standard OLS MM by using a variant (only the estimation parameters change, however,
34 This motivated the choice of 150 days estimation period prior to 30 June 2007 (before the GFC) to further examine the impact of the LB bankruptcy on the performance of 10 sectors and 4 financial industries (this is further discussed in the next paragraph).
62
the procedure for computing abnormal returns remain the same as discussed earlier
under the standard OLS MM) of the standard OLS MM to further examine the impact of
LB bankruptcy on the performance of sectors and financial industries in the following
ways. (1) Similar to Raddatz (2010), this study uses 150 days estimation period prior to
30 June 2007 to forecast normal period returns that would have eventuated in the
absence of GFC and LB bankruptcy. Why before 30 June 2007? The GFC started in
August 2007 (Dumontaux and Pop 2009). However, in July 2007, two of Bear Sterns
hedge funds specializing in subprime debt were filed for Chapter 15 bankruptcy and
Countrywide Financial, which was one of the biggest mortgage originators in the US,
notified about “difficult conditions” (Raddatz 2010). Therefore, for the purpose of this
study, the period before 30 June 2007 is assumed to be normal period. (2) In order to
control for multiple negative announcements during LB bankruptcy impairing the
variance estimates of AAR, a smaller event window of 5 trading days (−4 to 0 days)35 is
chosen as opposed to the typical lengths of 21 to 121 days (Peterson 1989). Two such
negative news announcements were the placement of Fannie Mae and Freddie Mac into
government conservatorship by the Federal Housing Finance Agency and the bailout of
the American International Group (AIG). Fannie Mae and Freddie Mac were put into
government conservatorship by the Federal Housing Finance Agency on Sunday, 7
September 2008 (Bartram and Bodnar 2009) and the subsequent day was 5 trading days
prior to the LB bankruptcy. On the other hand, AIG was bailed out on the day
subsequent to the event date (Bartram and Bodnar 2009). In order to control for the
aforementioned negative news announcements, 5 trading days prior to LB bankruptcy
and the days subsequent to the event date are not included in the event window.
According to the timeline of events provided in Bartram and Bodnar (2009)36, negative
news announcements on LB starts from 4 trading days prior to LB bankruptcy (9
September 2008). On −4 trading days (9 September 2008) ‘‘Lehman Brothers shares
plummet to lowest level on Wall Street in more than a decade’’ (Bartram and Bodnar
2009)37. On −3 trading days (10 September 2008), ‘‘Lehman Brothers puts itself up for
359 September 2008 to 15 September 2008. 36 For a detailed timeline of events that occurred during the GFC, refer to Bartram and Bodnar (2009). 37 The sharp decline in LB shares indicates that after Fannie Mae and Freddie Mac were put into government conservatorship by the Federal Housing Finance Agency on Sunday, 7 September 2008
63
sale after reporting a $4 billion loss and says it will spin off its troubled commercial real
estate assets’’ (Bartram and Bodnar 2009). On −1 trading days (12 September 2008),
‘‘with Lehman Brothers facing collapse, US officials struggle to find a buyer for the
distressed investment bank’’ (Bartram and Bodnar 2009). On the event day, ‘‘12.30am
EST: Lehman Brothers Holdings Incorporated files for Chapter 11 bankruptcy
protection. SEC Filing’’ (Bartram and Bodnar 2009). Since negative news
announcements on LB dominated the timeline of events for the period −4 trading days to
the event date (Bartram and Bodnar 2009), the impact of the LB bankruptcy is analyzed
for an event window of [−4, 0] trading days. In line with conventional event study
reporting38, this study focuses on the evolution of CAAR over the event window of [−4,
0] trading days. However, the daily AAR over the [−4, 0] trading days are also reported.
In sum, this study further examines the impact of LB bankruptcy on the performance of
10 sectors and 4 financial industries for an event window of [−4, 0] trading days (9
September 2008 to 15 September 2008) and an estimation period of 150 days prior to 30
June 2007 (22 November 2006 to 29 June 2007). The S&P 500 index is used as the
market index ( ). Finally, the standardized cross sectional test (discussed earlier) is
used for significance testing.
(Bartram and Bodnar 2009), the market participants anticipated that LB would be the next troubled investment bank, which greatly affected the investor confidence amongst LB shareholders. 38 ‘‘The event study literature typically privileges the analysis of cumulative returns because the
cumulative impact of the events is easier to visualize’’ (Raddatz 2010).
64
3.3 ROBUSTNESS TESTS FOR THE EVENT STUDY RESULTS
The market adjusted return model is employed as a robustness test to estimate abnormal
returns for each of the 10 sectors and 4 financial industries during the event window [−4,
0] trading days. Why? (1) Typically the event study approaches rely on past returns
immediately prior to the event period (estimation period) to compute expected returns
(returns that would have eventuated in the absence of the event under investigation). As
mentioned in the earlier section, since the GFC was a period characterized by high levels
of volatility, the use of an estimation period during the GFC will result in
misspecifications of alpha and beta parameters used for predicting normal period returns
during the event period. Due to the high levels of volatility, the estimated beta would be
much higher than what would have eventuated in normal circumstances. Nevertheless,
the market adjusted return model is used in scenarios whereby “it is not feasible39 to
have a pre-event estimation period for the normal model parameters” (MacKinlay 1997).
Accordingly, the market adjusted return model is used to check the robustness of this
study’s findings, which is based on 150 days estimation period prior to GFC, because it
does not depend on past returns to forecast normal period returns (Dumontaux and Pop
2009). (2) Brown and Warner (1985) find that, for daily data, the market adjusted return
model works reasonably well in examining the impact of an event and is also powerful
in the presence of event date clustering. Under the market adjusted return model, the
normal period return is the current event period market return. The statistical
significance of the AARs and CAARs are tested by employing the ordinary cross
sectional test approach (Brown and Warner 1985).
3.4 HYPOTHESIS DEVELOPMENT
Industry specific turmoil particularly arising from the financial sector is perceived to be
more dangerous and having a larger impact than country specific crises (Kenourgios et 39 Market adjusted return model is used for instance to investigate underpricing of initial public offerings
(Ritter 1991; MacKinlay 1997).
65
al. 2011). Bartram and Bodnar (2009) find that the GFC which emerged from the US
financial sector had adversely affected the stock performance globally, “and in nearly
every country, sector and industry”. In contrast, Hon et al. (2007) investigated whether
or not the impact of the 2000 Nasdaq technology bubble collapse in the US was
Technology, Media and Telecommunication (TMT) sector specific or pervasive in other
sectors40. Hon et al. (2007) find “that the collapse of the stock market in more than a
dozen countries is tied to close sectorial links (particularly in TMT) and cannot be
attributed to widespread contagion”. In other words, the collapse originated in the US
TMT sector and the effect from the collapse was limited to the TMT sector with sector
specific contagion effects in non US markets. As a result, the transmission effect was
TMT sector specific and not widespread on other sectors.
Accordingly, “contagion effects from a financial sector is perceived to be more
dangerous than in other industries because it arises very quickly, it spreads to a greater
extent within the industry, results in more failures and huge losses to creditors and
affects otherwise solvent financial institutions” (Kaufman 1994). In this study, it is
asserted that since the financial sector and the diversified financial industry were the
most exposed to LB, they will be the most significantly adversely affected from LB
bankruptcy. The following sub sections provide the theoretical foundation for the
propositions.
3.4.1 LB BANKRUPTCY AND THE US FINANCIAL SECTOR
The US financial sector is identified as “ground zero” and as the “epicenter” of the GFC
(Bartram and Bodnar 2009). Accordingly, Bartram and Bodnar (2009) investigated the
performance of world financial sector indices versus world non-financial sector indices
from 31 December 2006 to 27 February 2009. Bartram and Bodnar (2009) find the
world financial sector indices return being the most adversely affected (−63.9%) in
comparison to world non- financials (−38.3%) with volatility being almost 50% higher 40 Other sectors were namely; CD, CS, Eng, Fin, HC, Ind, IT, Mat, TS and Uti
66
than the non-financials. During the crisis period (12 September 2008 to 27 October
2008), there was a sharp drop in the returns for both the world financial sector (−40.5%)
and world non-financials (−35.7%) with increased standard deviations for each being
71.1% and 55.3% respectively. The US financial sector was also the worst performer in
terms of returns and having the most volatility when compared to that of Developed
markets excluding North America and EM. Furthermore, during the period 31 December
2006 to 27 February 2009, Bartram and Bodnar (2009) find the decline in the US index
return for the financial sector (−71.1%) higher than non-financials (−35.9%) with
standard deviations for each being 49.9% and 28.8% respectively. During the crisis
period the decline in return were quite similar with financial sector (−34.2%) being
slightly higher than non-financials (−32.3%) and each having standard deviations of
103.8% and 69.1% respectively.
Bartram and Bodnar (2009) also investigated the impact of the GFC (31 December 2006
to 27 February 2009 as full sample period) and the LB bankruptcy (12 September 2008
to 27 October 2008) on the US sector41 portfolio’s returns. The study looked at index
values return and standard deviation and revealed that all sectors suffered a decline in
returns with increased volatility during the entire sample period and also the crisis
period. During the entire sample period, the financial sector had the highest volatility
and was the most adversely affected. However, during the crisis period, Basic Materials
suffered the most declines in return and financials had the second highest volatility level
after Oil and Gas. Bartram and Bodnar (2009) also find the largest increase in
correlation across US sectors during the peak crisis period (15 September 2008 to 27
October 2008) in comparison to pre-crisis (1 January 2007 to 12 September 2008) and
post crisis periods (28 October 2008 to 27 February 2009). This study extends the work
of Bartram and Bodnar (2009) with the use of statistical methods (hypothesis testing,
event study approach and significance tests) and thoroughly examines the impact of LB
bankruptcy on the performance of NYSE sectors and financial industries.
41 The industries were namely, Oils and Gas, Basic Mat, Ind, Consumer Goods, HC, Consumer Services, TS, Uti, Fin and IT.
67
In sum, Bartram and Bodnar (2009) find all US sector indices suffering a decline in
returns and having increased volatility during LB bankruptcy. However, the US financial
sector had a higher standard deviation than the non – financials (Bartram and Bodnar
2009). LB being a constituent of the US financial sector is likely to associate more with
firms from the financial sector as opposed to firms from other sectors. Also, LB had
heavily invested in securities linked to the US sub-prime mortgage market. Financial
industries namely, banks, insurance and REITs were involved in subprime mortgages. It
follows that, since these financial industries are constituents of the financial sector, LB
was most associated with the financial sector in comparison to other sectors. In Table
3.2, correlation between LB price returns and each of the sectors index returns is
provided for the periods before the GFC (1994 to 2006) and during the GFC (2007 to 15
September 2008). Table 3.2 shows that the financial sector was significantly positively
correlated with LB at the 1% level. LB was most significantly correlated with the
financial sector (correlation coefficient was 0.63) during the GFC. This drives the
motivation for the need to examine whether the LB bankruptcy most significantly
adversely influenced the financial sector performance. The first set of null and alternate
hypotheses are stated below. H0 represents the null hypotheses while HA represents the
alternate hypotheses from here onwards.
H01: Financial sector performance is not the most significantly adversely affected during
LB bankruptcy.
HA1: Financial sector performance is the most significantly adversely affected during
LB bankruptcy.
68
Table 3.2: Correlation Results Between LB and each of the 10 sectors
Note: This table provides the correlation results between LB price returns and each of the sectors index returns for the periods before the GFC (1994 to 2006) and during the GFC (2007 to 15 September 2008). *, ** and *** represents significance at 10%, 5% and 1% levels respectively.
Periods Before the GFC (1994 to 2006) During the GFC (2007 to 15 September 2008)
FIN 0.09*** 0.63***
IND 0.14*** 0.08
CS 0.03** 0.1**
MAT 0.04* 0.14***
UTI 0.03 0.15***
TS 0.02 0.23***
IT 0.07*** 0.15***
HC 0.05** 0.14***
ENG 0.02 0.24***
CD 0.07*** 0.21***
3.4.2 LB BANKRUPTCY AND THE FINANCIAL INDUSTRIES
Acharya et al. (2009) mention LB bankruptcy having significant systematic risk
resulting in “the near collapse of the financial system”. Bartram and Bodnar (2009) find
US financial sector industries adversely affected during LB bankruptcy. Studies reveal
that during the GFC, subprime CDO market failure was a major driver in the
transmission of volatility to the large complex financial institutions (LCFIs) equity
returns (Longstaff 2010; Calice 2011). Calice (2011) states that “a distressed market for
ABS CDO can adversely affect the equity values of a large number of LCFIs. Such
critical effects have important stability implications for the banking system as a whole,
since a collapse in the equity prices for most banks would make it difficult for them to
raise common equity thus severely undermining the capital adequacy and solvency of
these institutions. This, in turn, could have important spillover effects on other LCFIs”
Calice (2011) motivated by 2007 to 2008 subprime crisis, examines the causal
relationship between subprime asset backed CDO market and the value of LCFIs in the
69
equity market using Asset Backed Securities (ABX) index returns and bank’s equity
returns. All LCFIs equity returns were found to be positively correlated with ABX index
returns. The study also finds volatility of ABX returns being transmitted to the equity
returns of the financial institutions. The findings of Calice (2011) support Longstaff
(2010) and confirm that during the crisis, subprime CDO market failure was a major
driver in the transmission of volatility to the equity prices of LCFIs. Both the US and the
European LCFIs were found to be prone to shocks from ABX. However, volatility in
ABX prices was found to lead the changes for European LCFIs equity price. On the
other hand, a two way causal relationship amongst the US LCFIs and ABX index was
found (Calice 2011).
Eichler et al. (2011) estimated banking crisis risk in the US (for major banks) during the
GFC in terms of short - term, long-term and total crisis risk. The study finds that the
escalating total crisis risk in the entire system was primarily influenced by short term
risk that commenced in summer 2007. During the crisis, non surviving banks (banks
that defaulted or were overtaken) suffered both long term and short term risks.
Conversely, surviving banks experienced fairly short term liquidity problems. The
results implied that problems in the mortgage market affected long term perspectives
about uncertainties in solvency which triggered the liquidity issue. Therefore, the
banking crisis resulted from illiquidity rather than solvency problems. “This liquidity
issue was the major problem (for the majority of banks), which was only triggered by
the doubts about solvency (because of problems in the mortgage market)”. As a result,
the long term risk is found to also increase during the crisis and particularly from 2008
onwards. Accordingly, the study supports the actions taken by the Federal Reserve and
the US government during the crisis in injecting liquidity and restoring confidence
“which finally saved the sound banks that were not conflicted by major solvency
problems” (Eichler et al. 2011).
70
Guo et al. (2011) investigated contagion effects within the stock market, real estate
market, credit default market and Energy market in both tranquil and turmoil periods
from October 2003 to March 2009 using Markov regime – switching VAR framework.
During the 2007 financial crisis, greater contagion effects were revealed amongst the
markets. The findings reveal that stock market activities (excluding its own shock)
triggered increased volatilities in the credit default market and energy markets.
However, the credit default market is less influenced by the housing market shock. Also,
contrary to the public’s view, real estate market was not significantly impacted by credit
default market and stock market (Guo et al. 2011).
Chakrabarty and Zhang (2010) tested competing theories of contagion by examining
how the shock from the LB bankruptcy impacted other firms disclosing whether they
had any credit and investment exposure to LB. The negative effects from contagion were
tested using microstructure variables (number, size and volume of trade, transaction
cost, price impact of trade, adverse selection cost and buy-sell order imbalance).
Chakrabarty and Zhang (2010) found that negative effects were experienced by firms
which were exposed to LB. This supports counterparty contagion hypothesis. The
negative effects were in the form of “greater price impact, information asymmetry,
selling pressure and negative returns”. Mixed evidence on the information asymmetry
hypothesis was found. Although firms disclosing non exposure to LB also suffered high
transaction costs, they did not experience greater “price impact, information asymmetry
and selling pressure”. Evidence supporting ‘flight to quality’ or ‘flight from liquidity’
was also not found (Chakrabarty and Zhang 2010).
Dumontaux and Pop (2009) examine the systematic nature of the collapse of LB on the
surviving financial institutions (“every institution operating in the same industry as LB
or in other fields of finance”) via contagion effect. In order for LB to be considered
systematically significant, its “failure should have significant adverse knock-on effects
on a large number of surviving financial institutions” (Dumontaux and Pop 2009). The
investigation was done from −2 to +2 days around LB bankruptcy. The empirical
71
findings of Dumontaux and Pop (2009) indicate that the impact of LB bankruptcy was
limited to the largest financial firms and it significantly affected surviving “non-bank”
financial services offering same products (mortgage and specialty finance, investment
services and diversified financial services firms) with similar financial conditions, risk
profile and characteristics. Further, collateral damages resulting from LB bankruptcy
were discriminatory towards the largest banks and financial institutions that were
apparently most exposed to LB (Dumontaux and Pop 2009). Therefore, the results
suggested that contagion effects were not generalized to all financial institutions but
were rather discriminatory towards biggest financial firms in the financial sector
(Dumontaux and Pop 2009).
Pichardo and Bacon (2009) investigated the impact of LB bankruptcy and tested the
market efficiency theory on 15 investment firms out of which about 9 had significant
stake in LB. Event study methodology was used with an event window of 30 days prior
and 30 days post the date of LB bankruptcy (15 September 2008). The CAAR was used
to test the market efficiency theory. The findings show that the 15 firms stock prices on
and around the event date was significantly negatively affected from LB bankruptcy.
This supported the semi-strong form of market efficiency. The results revealed that
approximately 24 days prior to the event date, the stock prices had started to decline.
This implies that the collapse of LB had been anticipated by the market (Pichardo and
Bacon 2009).
Raddatz (2010) investigated the impact of LB bankruptcy on stock price returns of 662
individual banks across 44 countries using an event study approach. The study tests
whether the differences in observed abnormal returns for the banks are a result of their
ex-ante reliance on wholesale funding. Raddatz (2010) revealed that globally and within
countries, those banks relying heavily on non-deposit sources of funds experienced a
substantial fall in stock returns. Also, liquidity was found to play a crucial role in
spreading the crisis. In comparison to banks with low wholesale dependence, the high
wholesale dependence ones suffered a decline in abnormal returns by greater than 2
72
percent within a country. Such a fall was observed during the three days subsequent to
the bankruptcy of LB. Further, Bartram and Bodnar (2009) find US financial industries
namely, financials, banks, insurance, Real Estate and Financial services adversely
affected with banks having the highest standard deviation during the period 12
September 2008 to 27 October 2008.
In sum, studies reveal that during the GFC, subprime CDO market was a major driver in
the transmission of volatility to the LCFIs equity returns (Longstaff 2010; Calice 2011).
Calice (2011) finds LCFIs (bank’s) equity returns being positively correlated with
subprime asset backed CDO market index returns. Raddatz (2010) finds that globally
and within countries, those banks relying heavily on non-deposit sources of funds
experienced a substantial fall in stock returns. Dumontaux and Pop (2009) find collateral
damages resulting from LB bankruptcy being discriminatory towards the largest banks
and the financial institutions that were apparently most exposed to LB. In support,
Pichardo and Bacon (2009) find the performance of investment firms (which are
constituents of the diversified financial industry), including those that had significant
stakes in LB, to be greatly adversely affected from the LB bankruptcy on and around the
event date (15 September 2008). Consequently, evidence suggests that the firms having
direct credit or investment exposures to LB suffered more than those not having direct
exposure to LB (Pichardo and Bacon 2009; Chakrabarty and Zhang 2010; Dumontaux
and Pop 2009). This is an indication of LB bankruptcy having a heterogeneous effect on
the US firm performance depending on their exposure to LB. This suggests that LB,
being an investment bank and a constituent of the diversified financial industry, would
have been more closely associated with other diversified financial firms than firms from
the other three financial industries (banks, insurances and REIT). This motivates the
investigation of whether the diversified financial industry amongst other financial
industries was the most significantly adversely affected from the LB bankruptcy. The
second set of null and alternate hypotheses are as follows.
H02: Diversified Financial industry’s performance was not the most significantly
adversely affected during LB bankruptcy.
73
HA2: Diversified Financial industry’s performance was the most significantly adversely
affected during LB bankruptcy.
3.5 CONCLUSION
In order to examine the performance of the US stock market during LB bankruptcy, the
S&P 500 index data has been disaggregated into 10 sectors and 4 financial industries
according to their respective GICS. In light of the idea of firm and sector heterogeneity
(Narayan and Sharma 2011), this study examines whether the financial sector and the
diversified financial industry were the most significantly adversely affected during LB
bankruptcy. Event study approach is used to analyze the impact of LB bankruptcy on the
performance of each of the 10 sectors and 4 financial industries. In sum, the following
two sets of null and alternate hypotheses have been set:
H01: Financial sector performance is not the most significantly adversely affected during
LB bankruptcy.
HA1: Financial sector performance is the most significantly adversely affected during
LB bankruptcy.
H02: Diversified financial industry’s performance is not the most significantly adversely
affected during LB bankruptcy.
HA2: Diversified financial industry’s performance is the most significantly adversely
affected during LB bankruptcy.
74
CHAPTER 4: DESCRIPTIVE STATISTICS RESULTS
4.0 INTRODUCTION
In this chapter, the descriptive statistics (mean, standard deviation, kurtosis and
skewness) of the sample used in this thesis is presented. There are three sections in this
chapter. In Section 4.1, descriptive statistics for the sample used in the standard OLS
MM are discussed. In the standard OLS MM, the estimation period ranges from 27
December 2007 to 31 July 2008 while the event window ranges from 01 August 2008 to
27 October 2008. In Section 4.2, descriptive statistics for the variables used in the
variant of the standard OLS MM are discussed. In the variant of the standard OLS MM,
the estimation period ranges from 22 November 2006 to 29 June 2007 while the event
window ranges from 9 September 2008 to 15 September 2008. Finally, the chapter
conclusion is discussed in Section 4.3.
4.1 DESCRIPTIVE STATISTICS FOR THE SAMPLE USED IN THE
STANDARD OLS MARKET MODEL
In Table 4.1, descriptive statistics for the sample used in the standard OLS MM are
presented. Table 4.1, Panel A reports mean, standard deviation, kurtosis and skewness
for the S&P 500 index return and each of the 10 sectors returns during the 150 days
estimation period immediately before the event window of [−30, +30] trading days. The
findings on mean returns show that the S&P 500 index and nine sectors (CD, CS, Fin,
HC, Ind, IT, Mat, TS, Uti) returns were adversely affected. The mean returns of the S&P
500 index, financials and TS sectors were the most adversely affected (−0.10%). The
financial sector has the highest standard deviation followed by the Eng sector. The
results on kurtosis show that the S&P 500 Index and all other sectors’ returns is
leptokurtic. In other words, they have kurtosis values greater than zero. The results on
skewness show that seven sectors (CS, Eng, HC, Ind, Mat, TS and Uti) had negative
skewness while the other three sectors (CD, Fin and IT) and the S&P 500 index had
positive skewness.
75
Table 4.1, Panel B reports mean, standard deviation, kurtosis and skewness for the S&P
500 index return and each of the 10 sectors returns during the event window of [−30,
+30] trading days. The findings on mean returns show that the S&P 500 index and nine
sectors (CD, CS, Eng, Fin, HC, Ind, Mat, TS, Uti) were adversely affected. The mean
returns of the Eng and Mat sectors were the most adversely affected (−1.03%). The
financial sector has the highest standard deviation followed by the Eng sector. The
results on kurtosis show that the S&P 500 Index and all other sectors’ returns is
leptokurtic. In other words, they have kurtosis values greater than zero. The skewness
results show that apart from CS and HC, the other eight sectors and the S&P 500 index
had positive skewness.
76
Table 4.1: Descriptive statistics for the sample used in the standard OLS MM
Note: This table provides descriptive statistics for the sample used in the standard OLS MM. Panel A provides descriptive statistics for the 150 days estimation period immediately before the event window of [-30, +30] trading days. The estimation period ranges from 27/12/2007 to 31/07/2008. Panel B provides descriptive statistics for an event window of [-30, +30] trading days. The event window ranges from 01/08/2008 to 27/10/2008.
Panel A: 150 days estimation period immediately before the event window [-30, +30] trading days
Mean Standard Deviation Kurtosis Skewness
S&P 500 Index -0.0010 0.0134 0.5658 0.2172 Consumer Discretionary -0.0006 0.0290 6.3796 0.2891
Consumer Staples -0.0007 0.0209 227.9000 -6.6273 Energy 0.0001 0.0310 23.3164 -1.2895 Financials -0.0010 0.0336 8.4733 0.8327 Health Care -0.0005 0.0226 83.6682 -3.4253 Industrials -0.0006 0.0255 48.3840 -1.8741 Information Technology -0.0006 0.0265 13.4095 0.2264
Materials -0.0001 0.0289 21.0045 -0.9713 Telecommunication Services -0.0010 0.0255 11.1300 -0.3021
Utility -0.0007 0.0150 1.6393 -0.0311
Panel B: Event window [-30, +30] trading days
Mean Standard Deviation Kurtosis Skewness
S&P 500 Index -0.0060 0.0343 2.3121 0.2228 Consumer Discretionary -0.0071 0.0442 2.8690 0.0649
Consumer Staples -0.0040 0.0320 4.8080 -0.0355 Energy -0.0103 0.0672 2.6066 0.3726 Financials -0.0053 0.0764 14.4802 0.9533 Health Care -0.0054 0.0383 17.5302 -0.1789 Industrials -0.0080 0.0420 3.6400 0.1938 Information Technology 0.0025 0.0307 23.3338 1.5971
Materials -0.0103 0.0556 3.9434 0.0107 Telecommunication Services -0.0020 0.0235 0.7703 0.2118
Utility -0.0049 0.0411 9.4696 0.6261
77
4.2 DESCRIPTIVE STATISTICS FOR THE VARIANT OF THE STANDARD
OLS MARKET MODEL
In Table 4.2, descriptive statistics for the sample used in the variant of the standard OLS
MM are presented. Table 4.2, Panel A reports mean, standard deviation, kurtosis and
skewness for the S&P 500 index return and each of the 10 sectors returns during the 150
days estimation period before 30 June 2007. The findings on mean returns show that
apart from the IT sector, the S&P 500 index and the other nine sectors (CD, CS, Eng,
Fin, HC, Ind, Mat, TS, Uti) had positive mean returns. The IT sector has the highest
standard deviation followed by Eng sector. The results on kurtosis show that the S&P
500 Index and all other sectors’ returns is leptokurtic. In other words, they have kurtosis
values greater than zero. The skewness results show that the S&P 500 index and all 10
sectors had negative skewness.
Table 4.1, Panel B reports mean, standard deviation, kurtosis and skewness for the S&P
500 index return and each of the 10 sectors returns during the event window of [−4, 0]
trading days. The findings on mean returns show that the S&P 500 index and nine
sectors (CD, CS, Eng, Fin, HC, Ind, Mat, TS, Uti) were adversely affected. The mean
return of the financial sector (−2.32%) was the most adversely affected followed by the
Eng sector (−1.565). The Eng and financial sectors respectively have the most standard
deviation. The results on kurtosis show that the S&P 500 Index and two sectors (Eng
and TS) return distribution are platykurtic (kurtosis values less than zero) while the other
eight sectors return distribution is leptokurtic. The skewness results show that apart from
HC and IT, the other eight sectors and the S&P 500 index had negative skewness.
78
Table 4.2: Descriptive statistics for the sample used in the variant of the standard OLS MM
Note: This table provides descriptive statistics for the sample used in the variant of the standard OLS MM. Panel A provides descriptive statistics for the 150 days estimation period before 30 June 2007. The estimation period ranges from 22/11/2006 to 29/06/2007. Panel B provides descriptive statistics for an event window [-4, 0] trading days. The event window ranges from 09/09/2008 to 15/09/2008.
Panel A: 150 days estimation period before the GFC Mean Standard
Deviation Kurtosis Skewness
S&P 500 Index 0.0005 0.0068 4.7566 -1.1974 Consumer Discretionary 0.0004 0.0185 243.8721 -8.0834 Consumer Staples 0.0004 0.0152 264.3636 -8.2661 Energy 0.0012 0.0206 171.4395 -7.1678 Financials 0.0002 0.0143 156.1416 -3.7306 Health Care 0.0005 0.0183 471.2135 -14.7086 Industrials 0.0008 0.0194 260.8024 -9.4978 Information Technology -0.0037 0.0285 14.2739 -0.6860 Materials 0.0014 0.0198 179.7449 -6.6381 Telecommunication Services 0.0009 0.0116 9.5136 -0.9501 Utility 0.0002 0.0143 440.8226 -13.7071 Panel B : Event window [-4, 0] trading days Mean Standard
Deviation Kurtosis Skewness
S&P 500 Index -0.0118 0.0270 -2.4272 -0.6267 Consumer Discretionary -0.0104 0.0267 1.5509 -0.4931 Consumer Staples -0.0026 0.0177 1.4374 -0.1860 Energy -0.0156 0.0677 -0.8684 -0.3816 Financials -0.0232 0.0546 34.6443 -3.8615 Health Care -0.0042 0.0217 4.6050 0.2676 Industrials -0.0070 0.0357 0.8132 -0.4047 Information Technology 0.0030 0.0302 10.7411 1.0494 Materials -0.0082 0.0512 0.8904 -0.8280 Telecommunication Services -0.0095 0.0208 -0.0606 -0.5261 Utility -0.0065 0.0313 6.0761 -1.7651
79
4.3 CONCLUSION
Overall the mean returns at the sector and the S&P 500 index level are found to be
adversely affected during the 150 days estimation period immediately before the event
window of [−30, +30] trading days (except for the Eng sector), event window of [−30,
+30] trading days (except for the Information Technology sector) and during the event
window of [−4, 0] trading days (except for the Information Technology sector). On the
other hand, the mean return at the sector and the S&P 500 index level during the 150
days estimation period before 30 June 2007 is generally found to be positive (except for
the Information Technology sector). In support, Park (2010) finds the mean return of
the S&P 500 index adversely affected after the global economic crisis while the mean
return of the S&P 500 index was found to be positively affected before the crisis.
The financial sector is found to have the most standard deviation during the 150 days
estimation period immediately before the event window of [-30, +30] trading days and
event window of [-30, +30] trading days. The Information Technology sector has the
highest standard deviation during the 150 days estimation period before 30 June 2007
while the Eng and Financial sectors respectively have the most standard deviations
during the event window of [−4, 0] trading days. Apart from the event window of [−4, 0]
trading days, the distribution of returns is generally found to be leptokurtic. During the
event window of [−4, 0] trading days, the distribution of returns for the S&P 500 index,
the Eng sector and the TS Sector were platykurtic. The results on skewness are mixed
(either negatively skewed or positively skewed). It is noteworthy that during the
estimation period of 150 days before the GFC, the distribution of the S&P 500 index
return and all 10 sectors returns were negatively skewed. In support, Park (2010) finds
the distribution of returns for the S&P 500 index negatively skewed before the global
economic crisis. Overall, the descriptive results show that the returns are not normally
distributed. Therefore, in this thesis, the Boehmer et al. (1991) standardized cross-
sectional test is used for testing the significance of the event study results.
80
CHAPTER 5: EVENT STUDY RESULTS FOR 10 SECTORS BASED ON THE
STANDARD OLS MM
5.0 INTRODUCTION
In this chapter event study results for the 10 sectors based on the standard OLS MM,
which has an estimation period of 150 days immediately before the event window of
[−30, +30] trading days, are presented. In the table, t - stats is reported in the paranthesis
and *, ** and *** represents significance at 10%, 5% and 1% level respectively. The
Boehmer et al. (1991) standardized cross sectional test approach is used for significance
testing. A positive sign implies positive impact from LB bankruptcy while a negative
sign implies adverse impact from LB bankruptcy. Furthermore, the AAR and CAAR
results for the following multi day event windows of [−5, 0], [−4, 0], [−3, 0], [−2, 0],
[−1, 0], [0, +1], [0, +2], [0, +3], [0, +4], [0, +5], [−1, +1], [−2, +2], [−3, +3], [−4, +4],
[−5, +5] are presented.
There are two main Sections in this chapter. In Section 5.1, event study results for the 10
sectors based on the standard OLS MM are discussed and the chapter conclusion is
provided in Section 5.2.
81
5.1 RESULTS FOR 10 SECTORS BASED ON THE STANDARD MM
Findings of t-test on the analysis of daily AARs and CAARs per sector appear in Table
5.1 and 5.2 respectively. Contrary to the first priori belief, the AARs results show that
amongst the 10 sectors, the financial sector performance is the least significantly
adversely affected (has the least number of negative AAR) and the most significantly
positively affected (has the most number of positive AAR) during the event window of
[−30, +30] trading days. Therefore, having an estimation period immediately before the
event window is not a good estimation parameter to predict normal period returns during
LB bankruptcy42 . In support, Dumontaux and Pop (2009) provide evidence on the
disadvantages of using an estimation period immediately prior to the event period of
[−2, + 2] days during LB bankruptcy in computing abnormal returns.
A preliminary evaluation of the financial sector returns data shows that the beta during
the estimation period (150 days), which is immediately before the event period, is higher
(1.62) than the actual event period beta (1.41). The estimation period returns
immediately prior to the event period are highly contaminated by the volatility from
negative news announcements during the GFC (Dumontaux and Pop 2009). As a result,
the estimated beta value resulted in inaccurate predictions of expected returns.
Consequently, the biased estimates from the estimation period resulted in positive
abnormal returns for the financial sector.
Although the [−30, +30] trading days event window, which is used in the standard OLS
MM, falls within the typical lengths of 21 to 121 days (Peterson 1989) used for event
study purposes, it would be appropriate to choose a smaller event window to examine
the impact of LB bankruptcy. The [−30, +30] trading days event period contains
42 This motivated the choice of 150 days estimation period prior to 30 June 2007 (before the GFC) to further examine the impact of the LB bankruptcy on the performance of 10 sectors and 4 financial industries (this is further discussed in the next paragraph).
82
multiple negative news announcements43 and all these negative news announcements
were not on LB. Negative news announcements on LB starts from 4 trading days prior to
LB (Bartram and Bodnar 2009). The CAAR results (reported in Table 5.2) shows that
the financial sector, which was the most exposed to LB, was significantly adversely
affected on the multi day intervals of [−4, 0] and [−3, 0] trading days. Therefore, it is
more appropriate to use a shorter event window of [−4, 0] trading days than the longer
[−30, +30] trading days.
Finally, the findings provide evidence in support of sector heterogeneity (Narayan and
Sharma 2011). It follows that because sectors are heterogeneous; the impact of LB
bankruptcy was not significant on all the 10 sectors. The impact from LB bankruptcy
varied depending on the sectors’ association with LB. Amongst the 10 sectors, LB had
the most association with the financial sector. Consequently, the financial sector was the
most affected from LB bankruptcy.
43 See Bartram and Bodnar (2009) for a timeline of events that occurred during the GFC.
83
Table 5.1: Findings of t-tests on the analysis of daily AARs per sector based on the standard OLS MM Note: This table reports AARs for 10 sectors based on the standard OLS MM with 150 days estimation period immediately before the event window of [-30, +30] trading days (01/08/2008 to 27/10/2008) around the event date (15 September 2008). t - stats is reported in the parentheses. *, ** and *** represents significance at 10%, 5% and 1% levels respectively based on the ordinary cross sectional approach for significance testing. A positive sign implies positive impact from the LB bankruptcy while a negative sign implies adverse impact from the LB bankruptcy.
Days Consumer Discretionary
Consumer Staples Energy Financials Health Care Industrials Information
Technology Materials Telecommunication Services Utilities
-30 0.0032 (1.5374)
0.0001 (-0.0432)
0.0019 (0.7556)
0.0115*** (3.7373)
-0.0088 (-1.6375)
0.0007 (0.4834)
0.0006 (0.1945)
-0.0112** (-2.1048)
-0.0036 (-1.206)
-0.0227*** (-8.6883)
-29 0.011*** (6.0463)
0.0144*** (7.7031)
-0.0592*** (-15.5173)
0.0037** (2.054)
0.0147*** (6.9961)
-0.0028 (0.1801)
0.0077** (2.5393)
-0.0244*** (-6.1304)
0.0076** (2.9779)
-0.0029 (0.0783)
-28 0.0128*** (6.0768)
0.0031 (0.7886)
-0.0218*** (-5.9637)
-0.0009 (-0.3962)
0.0079*** (3.0742)
-0.0041 (-0.5196)
-0.0075** (-2.3889)
-0.0182** (-2.1233)
-0.0302*** (-4.4544)
0.002 (1.1173)
-27 -0.009*** (-3.3372)
-0.0005 (0.494)
0.0287*** (7.3197)
-0.0098*** (-4.1119)
-0.0013 (-0.5365)
-0.0017 (-0.5775)
0.0077 (1.3536)
0.0084 (0.9109)
-0.0243*** (-20.1326)
-0.0038* (-1.8322)
-26 0.0017 (-0.0579)
-0.0074*** (-3.0102)
-0.0089*** (-3.7045)
-0.0073*** (-2.7766)
-0.0027 (-0.9852)
0.0061*** (2.9677)
0.0179*** (5.5482)
0.0039 (0.6092)
0.0114** (2.7612)
0.004 (1.6265)
-25 0.0183*** (7.1951)
0.0098*** (2.7504)
-0.0358*** (-7.8372)
0.0008 (1.2378)
0.0114*** (5.3274)
0.0036*** (3.1607)
-0.0025 (-0.9335)
-0.0177* (-1.7416)
-0.0501*** (-12.0695)
0.0035* (1.9158)
-24 0.0165*** (5.8988)
0.003 (0.6386)
-0.0119*** (-4.563)
0.0073*** (3.3494)
-0.0014** (-2.0403)
-0.0036 (-1.0358)
0.0039 (1.2567)
-0.016 (-1.6324)
0.0035 (1.2047)
0.0081*** (3.799)
-23 -0.0025 (-0.9309)
0.0091*** (5.3929)
0.0088** (2.5587)
-0.0226*** (-11.9004)
0.0027* (1.8266)
0.0029** (2.0685)
0.0083*** (5.6702)
0.016*** (4.9109)
-0.0154* (-2.2756)
-0.0117*** (-2.8168)
-22 -0.0137*** (-6.9042)
-0.0042** (-2.5374)
0.0465*** (21.3537)
-0.0185*** (-9.5826)
-0.0017 (-1.3766)
0.0014 (-0.5552)
0.0023 (0.8605)
0.0275*** (4.7023)
-0.0127 (-1.0879)
0.0093*** (3.5214)
-21 0.0149*** (6.9866)
0.006 (1.2545)
-0.021*** (-10.4136)
0.0154*** (6.7849)
0.0033 (1.2538)
0.0006 (-0.0378)
0.005* (1.9782)
-0.0072** (-2.1116)
0.0084** (3.0626)
-0.0148*** (-10.7503)
-20 0.0026* (1.6757)
0.0038*** (2.7637)
-0.0215*** (-7.9888)
0.0048*** (2.7163)
0.0074*** (4.4389)
0.001 (1.4793)
0 (0.0723)
-0.0088 (-0.9847)
-0.0198*** (-4.4648)
0.0061*** (3.0427)
-19 -0.0033 (-1.0701)
-0.0048 (-1.3564)
-0.006** (-2.3535)
-0.0077*** (-5.5535)
-0.0009 (0.15)
0.0026 (1.5902)
0.0004 (-0.2445)
0.0085** (2.5064)
0.0297*** (10.992)
0.0113*** (6.5368)
-18 -0.0148*** (-9.6814)
-0.0036*** (-2.9941)
0.0416*** (18.3021)
-0.012*** (-5.9899)
-0.0024 (-1.082)
-0.006*** (-5.7654)
-0.0059*** (-3.8715)
0.0049 (0.0917)
-0.0093** (-3.4609)
0.0057*** (4.3415)
-17 -0.0142*** (-7.9855)
-0.0035** (-2.4226)
0.0302*** (12.1418)
-0.0012 (-1.2226)
-0.0102*** (-6.2459)
-0.0064*** (-6.3193)
-0.0049** (-2.0797)
0.0107* (1.8073)
-0.0074** (-2.496)
0.0017 (0.6064)
-16 0.0013 (0.6001)
-0.0084*** (-4.0609)
0.0147*** (4.5831)
-0.0141*** (-9.5013)
0.0017 (1.0981)
-0.0018 (-1.6684)
-0.011*** (-3.5062)
0.0042 (1.2123)
-0.0176** (-3.1942)
0.0051** (2.4214)
-15 0.0083*** (6.2841)
0 (0.5247)
-0.0349*** (-19.8634)
0.0125*** (8.0442)
0.0025 (1.6146)
0.0012 (1.4322)
0.0001 (0.5229)
-0.0143*** (-4.9829)
-0.0411*** (-6.3337)
-0.0016 (-0.5942)
-14 -0.0013 (-1.0614)
-0.0021** (-2.1573)
0.0048** (2.346)
0.0017 (0.9212)
-0.0073*** (-4.1648)
-0.0017 (-1.3053)
0.0021 (0.9609)
0.003 (0.4672)
-0.0013 (0.2396)
0.0001 (0.2111)
-13 -0.0074*** (-2.793)
-0.0069*** (-3.6226)
0.0203*** (8.085)
0.0013 (0.7842)
-0.0011 (-1.2505)
-0.0033** (-2.6123)
-0.0096*** (-6.3942)
0.0038* (1.7694)
0.0046 (1.1966)
0.0065*** (6.5483)
84
Table 5.1 Findings of t-tests on the analysis of daily AARs per sector based on the standard OLS MM Continued ... Days Consumer
Discretionary Consumer
Staples Energy Financials Health Care Industrials Information Technology Materials Telecommunication
Services Utilities
-12 -0.0032** (-2.5829)
0.0014 (1.1992)
0.0125*** (4.9152)
0.001 (0.9482)
-0.0051*** (-4.0241)
0.0012 (-0.0841)
0.0021* (1.8799)
0.0052 (1.6484)
0.0001 (-0.3945)
0.0037* (1.9181)
-11 0.0068*** (4.1162)
-0.0029 (-0.9717)
-0.0283*** (-12.5992)
0.0097*** (5.5945)
0.0011 (0.3815)
0.0057*** (4.3577)
-0.0053*** (-4.0738)
0.0036* (1.7372)
-0.0182* (-2.2769)
-0.0015 (-0.6862)
-10 0.0094*** (5.3545)
-0.0031*** (-3.0483)
-0.0001 (-0.4157)
0.0178*** (11.8506)
-0.0027*** (-2.8308)
-0.0001 (-0.4122)
-0.0071*** (-3.1261)
0.0035 (0.501)
0.0274*** (4.2484)
-0.0087*** (-7.6299)
-9 0.0214*** (9.3056)
0.0119*** (5.4379)
-0.0615*** (-19.7706)
0.0229*** (9.4214)
0.003* (1.9946)
-0.0008 (0.8979)
-0.005** (-2.2979)
-0.0182 (-1.3105)
0.0245*** (3.7972)
-0.0097*** (-3.9436)
-8 0.0113*** (4.3975)
-0.0005 (-0.5794)
-0.0107* (-1.9137)
0.0136*** (8.2743)
0.001 (0.1053)
-0.0131*** (-2.9167)
-0.0183*** (-6.3123)
-0.0029 (0.2361)
-0.0029 (-0.9816)
-0.0096*** (-4.7578)
-7 0.0053 (1.4303)
0.0026** (2.2136)
0.0116*** (3.2753)
0.0092*** (5.8966)
-0.0089*** (-3.6621)
-0.0028 (-1.1892)
-0.0008 (-0.1947)
-0.0097** (-2.4347)
0.0529*** (5.9957)
0.0055*** (3.6696)
-6 -0.0004 (-1.0516)
0.0069*** (3.1654)
-0.0016 (-1.6129)
0.0175*** (5.3609)
-0.0091*** (-5.1759)
-0.0037** (-2.0494)
0.0023 (0.5406)
0.0079** (2.39)
0.0056 (0.7943)
-0.0142*** (-5.5986)
-5 0.0135*** (5.714)
0.0046** (2.3813)
-0.0407*** (-8.8665)
0.0131*** (5.4114)
0.0066*** (2.8729)
-0.0096* (-1.725)
-0.0174*** (-5.3724)
-0.0294*** (-3.6267)
-0.0204*** (-5.504)
0.0096*** (3.9977)
-4 0.0124*** (5.7564)
0.0135*** (5.3872)
-0.059*** (-14.3043)
0.0007 (0.1566)
0.0027 (1.4892)
-0.0015 (0.473)
0.0041* (1.7226)
-0.0157 (-1.1124)
-0.0101** (-2.4563)
-0.0083 (-1.1006)
-3 -0.0065*** (-3.8461)
-0.004** (-2.5729)
0.0371*** (9.2338)
-0.0142*** (-4.9861)
0.0005 (-0.5756)
0.0053 (0.6692)
0.0035 (1.0528)
0.0157*** (2.813)
-0.012** (-2.8809)
0.0054 (1.1695)
-2 -0.0043*** (-2.8153)
0.0027 (1.2502)
0.0058 (1.0842)
-0.0097*** (-4.3193)
0.0068 (1.5022)
0.0043 (1.2385)
-0.0111*** (-3.5664)
0.0003 (0.5278)
-0.0122** (-2.8318)
0 (0.2129)
-1 -0.011*** (-4.8071)
0.0009 (0.6011)
0.0357*** (12.4307)
-0.0013 (-0.7252)
-0.0019 (-1.4773)
0.0072*** (2.9309)
-0.0014 (-0.871)
0.0285*** (7.5631)
0.0009 (-0.2542)
0.0127*** (5.7607)
0 0.0283*** (10.1184)
0.0116*** (5.799)
-0.0515*** (-11.9887)
0.003 (0.5626)
0.0047*** (3.736)
0.0107*** (3.9481)
0.0112*** (4.8398)
-0.0011 (1.5998)
0.0455*** (6.4051)
-0.0086 (-0.7254)
1 -0.0122*** (-5.6322)
-0.0104*** (-4.0072)
0.0169*** (5.7417)
0.0161*** (2.8207)
-0.012*** (-3.7788)
-0.0059** (-2.315)
-0.0125*** (-3.076)
-0.005 (-0.7304)
-0.0019 (-0.7301)
-0.0238** (-2.2349)
2 0.0076*** (2.8213)
-0.0016 (0.4971)
0.0151** (2.4997)
-0.0031 (-1.2688)
-0.0083*** (-2.6853)
0.0062 (1.53)
0.0137* (1.7731)
0.0061 (0.9386)
0.0639*** (5.2892)
-0.0267*** (-5.2667)
3 -0.0182*** (-3.5225)
-0.0045** (-2.1387)
-0.0018 (-0.3339)
0.0363*** (3.5302)
0.0024 (0.7882)
-0.0112** (-2.4501)
-0.0094* (-1.8101)
-0.0197*** (-2.926)
-0.0509*** (-5.3789)
0.0106** (2.5473)
4 -0.0265*** (-5.2693)
-0.0143*** (-3.5439)
0.0517*** (9.0917)
0.0301* (1.892)
-0.0122*** (-4.6256)
-0.0148*** (-3.6815)
-0.0127*** (-3.3329)
0.0126 (1.2998)
-0.025*** (-3.8129)
0.0064 (0.1792)
5 -0.0044 (-1.6133)
-0.0017 (-0.7116)
0.0179*** (4.7087)
-0.0216*** (-3.8135)
-0.0041 (-0.9919)
0.0024 (1.1984)
-0.0028 (-0.8656)
0.0086 (0.7583)
0.0049 (0.2351)
0 (-0.1116)
6 0.0049** (2.1995)
0.0052** (2.402)
-0.0194*** (-4.0494)
0.0097*** (2.7375)
0.0048*** (3.1616)
-0.0053*** (-2.8071)
0.0042*** (2.8229)
-0.0109 (-1.6847)
0.0041 (0.2629)
-0.0039* (-2.0235)
7 -0.0005 (-1.188)
-0.0006 (0.2789)
-0.0021 (-0.1824)
-0.0097** (-2.0068)
0.0043** (2.0391)
-0.0079*** (-3.9724)
0.0078*** (3.0249)
-0.0055 (-0.9015)
-0.0095* (-2.1339)
0.0027** (2.3889)
85
Table 5.1 Findings of t-tests on the analysis of daily AARs per sector based on the standard OLS MM Continued ... Days Consumer
Discretionary Consumer
Staples Energy Financials Health Care Industrials Information Technology Materials Telecommunication
Services Utilities
8 -0.0104*** (-2.8264)
0.0056** (2.2556)
0.0007 (0.8381)
-0.0111** (-2.0532)
0.0025 (0.8707)
-0.0157*** (-6.286)
-0.0089*** (-2.79)
-0.0163* (-1.8754)
-0.0078 (-1.5305)
0.012*** (6.1426)
9 0.0003 (0.206)
0.0029 (1.1469)
-0.0339*** (-13.8713)
0.0117** (2.1462)
0.0002 (-0.2748)
-0.0125*** (-3.7549)
-0.002 (-0.5063)
-0.03*** (-5.1095)
0.0181** (2.6719)
-0.0103*** (-4.3202)
10 0.0398*** (9.1455)
0.0134*** (5.4761)
-0.0539*** (-9.5876)
-0.0016 (0.3592)
-0.0078 (-0.9912)
0.0272*** (7.6655)
0.007* (1.7332)
0.0081** (2.6795)
0.0854*** (7.8012)
0.0005 (1.0719)
11 -0.0382*** (-8.5245)
-0.0143*** (-4.3133)
0.0185*** (3.3381)
0.029** (2.1289)
0.0003 (-0.2507)
-0.0202*** (-5.3912)
-0.0072** (-2.1863)
-0.0226*** (-5.8042)
-0.0613*** (-7.4871)
-0.0166*** (-5.8083)
12 -0.0024 (-1.0026)
0.0159*** (4.8552)
-0.0336*** (-8.5575)
0.0119 (-0.2874)
-0.0042 (-1.2731)
-0.0214*** (-6.7912)
-0.0137*** (-4.3036)
-0.0196*** (-3.376)
0.0186** (2.7178)
0.008*** (3.1452)
13 0.0015 (0.0945)
0.0072*** (2.8969)
-0.0691*** (-11.7682)
0.0112 (0.4339)
-0.0006 (0.6358)
-0.0196*** (-4.0979)
-0.0052 (-1.4657)
-0.0408*** (-3.4412)
0.0062 (1.1646)
0.0018 (0.6601)
14 -0.0176*** (-6.9697)
-0.0061* (-1.9383)
0.0036 (0.7618)
0.0014 (-0.4105)
-0.0007 (-0.1294)
0.0009 (0.381)
-0.0049 (-1.4383)
0.0134* (1.9409)
0.0111** (2.4126)
-0.0039 (-1.3882)
15 0.0153*** (3.9785)
-0.0054 (-0.9888)
-0.0303*** (-5.2095)
0.0228*** (3.3034)
-0.0382*** (-10.7622)
0.0109** (2.5316)
0.0014 (-0.0685)
0.0054* (1.7456)
0.0298*** (5.9724)
-0.0313*** (-7.1553)
16 0.0094** (2.4058)
0.0109** (2.1715)
-0.034*** (-5.3264)
0.0056 (0.5248)
0.0036 (1.3853)
0.0091** (2.3753)
0.0028 (0.6164)
-0.0052 (-0.1726)
0.0671*** (8.2807)
-0.0094 (-1.3878)
17 0.003 (0.4556)
-0.0135*** (-3.6079)
0.0148*** (2.7612)
-0.0242*** (-2.9079)
-0.0084** (-2.1457)
0.007 (0.6151)
0.0098** (2.4395)
0.0351*** (3.2159)
-0.012*** (-3.6251)
-0.0126*** (-3.4916)
18 0.016*** (3.0585)
-0.0151*** (-3.3079)
-0.0353*** (-5.5444)
-0.0093 (-1.242)
-0.0279*** (-6.3981)
0.0221*** (4.1574)
0.0472*** (10.844)
0.0107* (2.0229)
0.0628*** (7.4007)
-0.023*** (-5.599)
19 0.0099 (0.6271)
-0.0162*** (-3.3734)
-0.0686*** (-9.2447)
0.088*** (7.0574)
-0.0276*** (-6.5199)
0.0015 (0.0531)
-0.0013 (-0.5126)
0.0037 (0.2756)
-0.0122*** (-4.1328)
-0.0462*** (-8.5196)
20 -0.0596*** (-5.1505)
0.0057 (1.0576)
0.1191*** (11.4171)
-0.0648*** (-2.9461)
0.0639*** (8.0351)
-0.0119* (-1.6963)
-0.0137*** (-2.8482)
-0.0013 (-0.7312)
-0.0671*** (-5.9163)
0.0723*** (9.8311)
21 -0.0214*** (-4.9726)
-0.0145** (-2.4851)
0.0075 (1.3671)
0.0712*** (4.3061)
0.0024 (0.282)
-0.0149*** (-3.1012)
-0.0275*** (-6.088)
-0.0195** (-2.6741)
-0.0138*** (-5.3992)
-0.0064 (-1.5795)
22 0.017*** (2.6993)
-0.0025 (-0.9171)
-0.0959*** (-15.7732)
0.0399*** (4.8914)
-0.0248*** (-4.9908)
-0.0057 (-1.2617)
0.0116** (2.1706)
-0.0296*** (-5.2307)
0.0838*** (7.4262)
-0.0288*** (-7.9732)
23 -0.0097 (-1.6099)
0.0097* (2.0048)
0.0327*** (3.4515)
-0.0474*** (-7.1341)
0.0067 (1.4551)
-0.0031 (-0.7659)
-0.0013 (-0.1467)
0.007 (0.8049)
-0.0561*** (-6.572)
0.0183*** (3.7173)
24 0.0092** (2.2316)
0.0105* (1.8942)
0.0235*** (2.8195)
0.0002 (0.0566)
0.0048 (0.8128)
-0.0052** (-2.2389)
0.0016 (0.2172)
0.0013 (-0.3934)
0.0127** (2.4343)
0.0102** (2.1109)
25 -0.0331*** (-6.4542)
-0.0051 (-1.005)
0.0766*** (11.998)
-0.0482*** (-5.6416)
0.0103** (2.5314)
-0.0105*** (-2.8163)
-0.0145*** (-3.4917)
0.0302*** (4.6844)
-0.0072 (-1.2586)
0.0667*** (10.515)
26 0.0092** (2.2169)
-0.0009 (0.3781)
-0.0122** (-2.6866)
0.0296*** (5.7847)
-0.0008 (0.1118)
-0.0075 (-1.4341)
-0.0223*** (-6.3472)
-0.0148*** (-3.2365)
0.0075 (1.6656)
-0.0153*** (-5.62)
27 0.0133** (2.0386)
0.0006 (-0.1403)
-0.0734*** (-10.9353)
0.0214** (2.4868)
-0.0217** (-2.6102)
0.0054 (1.405)
0.0203** (2.4086)
-0.0205** (-2.383)
0.0759*** (6.4536)
-0.0221*** (-5.4303)
86
Table 5.1 Findings of t-tests on the analysis of daily AARs per sector based on the standard OLS MM Continued ... Days Consumer
Discretionary Consumer
Staples Energy Financials Health Care Industrials Information Technology Materials Telecommunication
Services Utilities
28 -0.04*** (-7.3902)
-0.0026 (-0.5191)
0.0222*** (3.2032)
-0.0484*** (-6.0148)
-0.015* (-1.985)
-0.0224*** (-3.613)
-0.0288*** (-5.3934)
-0.0409*** (-3.6675)
0.0164*** (4.8236)
0.0319*** (6.4316)
29 0.0017 (-0.1631)
-0.0164*** (-5.9652)
-0.0146*** (-3.6147)
0.0248*** (3.8408)
-0.0077** (-2.6794)
-0.0042 (-1.6024)
0.0153*** (3.1464)
0.0094 (0.7029)
0.0378*** (5.0749)
-0.012*** (-3.9344)
30 0.0112** (2.1406)
-0.0018 (-0.3271)
-0.0454*** (-8.7265)
0.0081 (-0.3654)
-0.0276*** (-4.3688)
0.0045 (0.9529)
0.0131*** (2.9735)
-0.0142*** (-3.0041)
0.0147*** (3.907)
-0.0171*** (-5.092)
87
Table 5.2 Findings of t-tests on the analysis of CAARs per sector based on the standard OLS MM Note: This table reports CAARs for 10 sectors based on the standard OLS MM with 150 days estimation period prior to the event period. t - stats is reported in the parentheses. *, ** and *** represents significance at 10%, 5% and 1% levels respectively based on the standardized cross sectional approach for significance testing. A positive sign implies positive impact from LB bankruptcy while a negative sign implies adverse impact from LB bankruptcy.
Window Consumer Discretionary
Consumer Staples Energy Financials Health Care Industrials Information
Technology Materials Telecommunication Services Utilities
[-5, 0] 0.0324*** (5.6333)
0.0292*** (5.2254)
-0.0726*** (-6.6579)
-0.0083 (-0.5163)
0.0195*** (3.7496)
0.0164*** (3.2238)
-0.011 (-1.2326)
-0.0016 (1.555)
-0.0082 (-1.0257)
0.0108* (1.9375)
[-4, 0] 0.0189*** (3.3189)
0.0247*** (5.5778)
-0.0319*** (-3.6422)
-0.0215* (-1.7011)
0.0129** (2.2098)
0.0259*** (4.0516)
0.0064 (1.4326)
0.0278*** (4.5479)
0.0122 (1.5827)
0.0012 (0.913)
[-3, 0] 0.0065 (0.7783)
0.0112*** (3.0634)
0.0271*** (4.164)
-0.0222* (-1.9491)
0.0101* (1.7573)
0.0275*** (4.1024)
0.0023 (0.712)
0.0434*** (7.5569)
0.0223** (2.8683)
0.0095* (2.0091)
[-2, 0] 0.013** (2.6219)
0.0152*** (4.233)
-0.01 (-1.6146)
-0.008 (-0.903)
0.0096** (2.5234)
0.0222*** (4.5587)
-0.0013 (0.0927)
0.0277*** (5.7225)
0.0342*** (4.7854)
0.0041 (1.3985)
[-1, 0] 0.0173*** (4.82)
0.0125*** (4.2468)
-0.0158*** (-2.7547)
0.0017 (0.0704)
0.0028** (2.0787)
0.0179*** (5.2266)
0.0098*** (3.3804)
0.0274*** (7.3627)
0.0464*** (6.2856)
0.0041* (1.7051)
[0, +1] 0.0161*** (4.0042)
0.0012 (0.6069)
-0.0346*** (-5.2206)
0.0191** (2.1173)
-0.0073 (-1.0912)
0.0048 (1.2416)
-0.0013 (0.0058)
-0.0061 (0.2256)
0.0436*** (5.6573)
-0.0323* (-1.8188)
[0, +2] 0.0237*** (4.3721)
-0.0003 (0.6811)
-0.0195** (-2.3948)
0.016 (0.9945)
-0.0156** (-2.6027)
0.0109 (1.6451)
0.0124 (1.5317)
0.0001 (0.6693)
0.1075*** (5.9347)
-0.059*** (-2.7793)
[0, +3] 0.0055 (0.248)
-0.0049 (-0.6654)
-0.0213** (-2.4017)
0.0523*** (2.6806)
-0.0132* (-1.8059)
-0.0003 (-0.0151)
0.003 (0.6958)
-0.0196 (-0.8573)
0.0566*** (3.8309)
-0.0484* (-1.8941)
[0, +4] -0.021*** (-4.0862)
-0.0192*** (-3.4919)
0.0304*** (4.1791)
0.0824*** (4.6674)
-0.0254*** (-5.5744)
-0.0151*** (-3.1755)
-0.0097 (-0.4541)
-0.007 (-0.3121)
0.0316** (2.758)
-0.042** (-2.1063)
[0, +5] -0.0254*** (-4.9203)
-0.0209*** (-3.86)
0.0483*** (5.6043)
0.0608*** (3.9668)
-0.0294*** (-5.1782)
-0.0126*** (-2.737)
-0.0125 (-0.6761)
0.0015 (-0.0345)
0.0365** (2.9361)
-0.042** (-2.4029)
[-5, +5] -0.0212*** (-2.9428)
-0.0033 (-0.4736)
0.0272** (2.5536)
0.0495** (2.4639)
-0.0147** (-2.5419)
-0.0069 (-1.0833)
-0.0346** (-2.414)
0.0011 (0.6078)
-0.0172 (-0.9145)
-0.0227 (-0.6499)
[-4, +4] -0.0304*** (-4.4346)
-0.0062 (-1.0773)
0.05*** (4.9157)
0.0579** (2.549)
-0.0172*** (-3.3143)
0.0002 (-0.7155)
-0.0145 (-0.6979)
0.0218 (1.641)
-0.0017 (0.2614)
-0.0322 (-1.2254)
[-3, +3] -0.0163*** (-2.8852)
-0.0053 (-0.5596)
0.0573*** (5.9646)
0.0272 (1.124)
-0.0078 (-1.4485)
0.0166 (1.4753)
-0.0059 (-0.0854)
0.0249 (1.6347)
0.0334** (2.7991)
-0.0303 (-1.0171)
[-2, +2] 0.0084 (0.6022)
0.0033 (1.3061)
0.022** (2.4796)
0.0051 (0.1249)
-0.0106* (-1.9877)
0.0224*** (2.9461)
0 (0.3296)
0.0289** (2.5989)
0.0962*** (5.9826)
-0.0463* (-1.969)
[-1, +1] 0.0051 (0.5168)
0.0021 (0.8807)
0.0011 (0.9334)
0.0178 (1.4484)
-0.0091* (-1.992)
0.012** (2.5702)
-0.0026 (-0.3932)
0.0224*** (3.1656)
0.0445*** (6.3527)
-0.0196 (-0.9283)
88
5.2 CONCLUSION
This chapter provides evidence that the standard OLS MM provides biased estimates to
predict normal period returns for the financial sector. Consequently, the biased
estimation parameters lead to an inaccurate positive abnormal return performance for the
financial sector. This motivates the use of a variant of the standard OLS MM, which has
an estimation period of 150 days before the GFC (before 30 June 2007), for further
analyses of the 10 sectors and 4 financial industries.
In the next two chapters, event study results for each of the 10 sectors and 4 financial
industries, which are based on the estimation period of 150 days before the GFC, are
presented and discussed respectively.
89
CHAPTER 6: EVENT STUDY RESULTS FOR THE 10 SECTORS
6.0 INTRODUCTION
In this chapter event study results for the 10 sectors are presented. The abnormal return
performances of all the 10 sectors were analyzed to determine whether the financial
sector, which was identified as the epicenter for the GFC (Bartram and Bodnar 2009),
was the most significantly adversely affected during LB bankruptcy. The event study
results are based on the variant of the standard OLS MM with an estimation period
before 30 June 2007 and the robustness test results are based on the market adjusted
return model. For each of the 10 sectors, the daily AAR is reported for a 5 day event
window (from −4 to 0 trading days44). In the table, t - stats is reported in the parentheses
and *, ** and *** represents significance at 10%, 5% and 1% levels respectively. The
Boehmer et al. (1991) standardized cross sectional test approach is used for significance
testing. A positive sign implies positive impact from the LB bankruptcy while a negative
sign implies adverse impact from the LB bankruptcy. Furthermore, the evolution of
CAAR results over the multi day intervals of [−4, −3], [−4, −2], [−4, −1] and [−4, 0] are
presented.
This chapter has three main sections. In Section 6.1, AAR and CAAR results for the 10
sectors, which are based on the OLS MM and an estimation period of 150 days prior to
30 June 2007, are discussed. Section 6.2 presents the robustness test results based on the
market adjusted return model. Finally, the chapter conclusion is provided in Section 6.3.
44 9 September 2008 to 15 September 2008
90
6.1 OLS MM RESULTS BASED ON AN ESTIMATION PERIOD OF 150 DAYS
BEFORE 30 JUNE 2007
In this section, event study results for the 10 sectors, which is based on the OLS MM
with 150 days estimation period prior to 30 June 2007 (22 November 2006 to 29 June
2007) and event window of [−4, 0] trading days (9 September 2008 to 15 September
2008) around the event date (15 September 2008), are presented. Findings of t-test on
the analysis of daily AARs and CAARs per sector appear in Tables 6.1 and 6.2
respectively. The main findings on the performance of the sectors on the event date and
over the multi day intervals leading to the event date, which is based on the AARs and
CAARs results, are summarized in the following paragraphs.
On the event date, all 10 sectors AARs are significantly affected. CD, CS, Eng, HC, Ind,
IT, TS and Uti sectors’ AAR are significantly affected at the 1% level while Mat and Fin
are significantly affected at the 5% and 10% levels respectively. Fin and Eng sectors
were the only two sectors that have negative AARs. Conversely, the other eight sectors
have positive AAR suggesting positive effects from LB bankruptcy. This implies that
since Fin and Eng were most correlated45 with LB, they were most adversely affected
from LB bankruptcy on the event date. The CAAR for the [−4, 0] trading days event
window show that seven sectors were significantly affected (CS, Fin, HC, Ind, IT, Mat
and Uti) while three were insignificantly affected (CD, Eng and TS). Amongst the
significantly affected sectors, LB bankruptcy had an adverse impact on only the Fin
sector (−4.77%). The AAR and CAAR results provide evidence in support of sector
heterogeneity (Narayan and Sharma 2011) by showing that not all sectors’ performance
were significantly adversely affected from the LB bankruptcy on the event date and
during the [−4, 0] trading days event window. During the LB bankruptcy, the returns of
sectors were affected differently.
45 During the GFC, Fin had the highest correlation coefficient of 0.63 followed by Eng having the second highest correlation coefficient of 0.24. Both correlation coefficients are significant at the 1% level.
91
The CAAR results show that the financial sector is the only sector consistently
significantly adversely affected over the multi day intervals of [−4, −3], [−4, −2], [−4,
−1] and [−4, 0] trading days. Consequently, the Fin sector has the most adverse CAAR
of −4.77%, significant at the 5% level during the event window of [−4, 0] trading days.
This implies that due to the strong degree of correlation between LB and the Fin sector,
negative news announcements on LB had an adverse impact on the Fin sector
performance. The financial sector is not only significantly adversely affected over the
event window of [−4, 0] trading days but also the effect of LB bankruptcy on the
financial sector is significantly different from all the other nine sectors (see Table 6.3).
Table 6.3 presents a comparison between the impact of LB bankruptcy on the financial
sector against each of the other nine sectors CAAR over the event window [-4, 0]
trading days. The results show that the impact from LB bankruptcy on the Financial
sector is significantly different from all the other sectors largely at the 1% level of
significance with the exception of Eng and TS sector being significantly different at the
5% level. Overall, the results provide evidence in support of the proposition that the
performance of the financial sector was the most significantly adversely affected from
the LB bankruptcy during the event window of [−4, 0] trading days.
Moreover, amongst the 10 sectors, only Fin sector’s adverse performance correlates
directly with negative news announcements on LB. According to the timeline of events
and policy actions for the financial crisis provided in Bartram and Bodnar (2009), the
trading days on which negative information were released included 9 September 2008,
10 September 2008, 12 September 2008 and 15 September 2008. These are −4, −3, −1
and 0 trading days respectively. On −4, −3 and 0 trading days, the AAR results for the
Fin sector are negative and also largely significantly affected. Generally, the CAAR
results (reported in Table 6.2) provide evidence in support of a semi strong form of
market efficiency by showing a direct relationship between the price return
performances of LB and the price return performance of the Fin sector in the following
ways. (1) Over the multi day intervals of [−4, −3], [−4, −2], [−4, −1] and [−4, 0] trading
days, Fin is the only sector with a consistent negative CAAR. (2) Overall, the Fin sector
92
is the only sector that was significantly adversely affected from LB bankruptcy over the
[−4, 0] trading days window. This is depicted by the negative CAAR for the [−4, 0]
multi day interval being significant at the 5% level.
In sum, this investigation depicts two important messages. Firstly, this study finds
evidence in support of sector heterogeneity (Narayan and Sharma 2011). It follows that
because sectors are heterogeneous; not all sectors’ performance was significantly
adversely affected from the LB bankruptcy on the event date and during the multi day
intervals leading to the event date. Amongst the 10 sectors, LB had the most association
with the financial sector (see Table 3.2). This resulted in the Fin sector performance not
only being significantly adversely affected on the event date but also being the most
significantly adversely affected during the multi day interval of [−4, 0] trading days
being the period over which most negative news on LB was released. Secondly, this
study finds evidence in support of a semi strong form of market efficiency. Since the Fin
sector was the most significantly correlated with LB, negative news announcements on
LB had a significantly adverse impact on the performance of the financial sector. This
implies that the US share market and investors instantly reacted to the negative news
announcements on LB.
93
Table 6.1: Findings of t-tests on the analysis of daily AARs per sector based on the variant of the standard OLS MM
Note: This table reports AARs for 10 sectors based on the OLS MM with 150 days estimation period prior to 30 June 2007 (22/11/2006 to 29/06/2007) and event window of [-4, 0] trading days (8/9/2008 to 22/9/2008) around the event date (15 September 2008). t - stats is reported in the parentheses. *, ** and *** represents significance at 10%, 5% and 1% levels respectively based on the Boehmer et al. (1991) standardized cross sectional approach for significance testing. A positive sign implies positive impact from the LB bankruptcy while a negative sign implies adverse impact from the LB bankruptcy.
Days Consumer Discretionary
Consumer Staples Energy Financials Health Care Industrials Information
Technology Materials Telecommunication Services Utilities
-4 0.0044** (2.2509)
0.0171*** (5.8937)
-0.0454*** (-9.6793)
-0.0161*** (-4.1848)
0.011*** (3.7345)
-0.0021 (-0.177)
0.0058*** (3.0516)
-0.0114 (-1.0267)
-0.0091** (-2.481)
0.0057** (2.1285)
-3 -0.0041* (-1.9854)
-0.0048*** (-2.9661)
0.0358*** (8.849)
-0.0104*** (-4.2938)
-0.0016 (-1.3918)
0.0056 (0.7414)
0.0038 (1.1381)
0.0139*** (3.045)
-0.0116 (-1.702)
0.003 (0.6302)
-2 -0.0001 (-0.4079)
0.0012 (0.5211)
0.0008 (-0.3431)
-0.0016 (0.2238)
0.003 (-0.4454)
0.0048* (1.9808)
-0.0109*** (-4.0864)
-0.0027 (0.0096)
-0.0134** (-2.7744)
-0.0055** (-2.2897)
-1 -0.0095*** (-4.368)
0.0005 (0.5786)
0.0357*** (11.0839)
0.0005 (-0.2667)
-0.0023 (-1.4721)
0.0074*** (2.7202)
-0.0006 (-0.5757)
0.0273*** (6.3863)
-0.0073 (-1.1481)
0.0119*** (5.3952)
0 0.0172*** (6.0565)
0.0166*** (6.157)
-0.0329*** (-6.0064)
-0.0202* (-1.8583)
0.0164*** (5.8127)
0.0099*** (3.2206)
0.0136*** (5.0996)
0.0051** (2.6253)
0.0478*** (6.5727)
0.0107*** (2.8311)
Table 6.2: Findings of t-tests on the analysis of CAARs per sector based on the variant of the standard OLS MM
Note: This table reports CAARs for 10 sectors based on the OLS MM with 150 days estimation period prior to 30 June 2007 (22/11/2006 to 29/06/2007) for various multi day intervals leading to the event date (15 September 2008). t - stats is reported in the paranthesis. *, ** and *** represents significance at 10%, 5% and 1% levels respectively based on the Boehmer et al. (1991) standardized cross sectional approach for significance testing. A positive sign implies positive impact from the LB bankruptcy while a negative sign implies adverse impact from the LB bankruptcy.
Days Consumer Discretionary
Consumer Staples Energy Financials Health Care Industrials Information
Technology Materials Telecommunication Services Utilities
-4 0.0044** (2.2509)
0.0171*** (5.8937)
-0.0454*** (-9.6793)
-0.0161*** (-4.1848)
0.0109*** (3.7345)
-0.0021 (-0.177)
0.0058*** (3.0516)
-0.0114 (-1.0267)
-0.0091** (-2.481)
0.0057** (2.1285)
[-4, -3] 0.0003 (0.3913)
0.0123*** (4.4256)
-0.0096 (-0.7275)
-0.0265*** (-5.0055)
0.0099* (2.0037)
0.0035 (0.4623)
0.0096*** (3.0373)
0.0025 (1.1655)
-0.0208** (-3.5389)
0.0086** (2.543)
[-4, -2] 0.0002 (0.1487)
0.0134*** (3.9056)
-0.0088 (-0.7992)
-0.0281*** (-4.4801)
0.0134 (1.2759)
0.0083 (1.5555)
-0.0013 (0.3407)
-0.0002 (0.9845)
-0.0342*** (-7.9963)
0.0031 (0.9636)
[-4, -1] -0.0093** (-2.1933)
0.014*** (3.5166)
0.0269*** (4.2467)
-0.0275** (-2.4326)
0.0111 (0.582)
0.0157*** (2.6615)
-0.0019 (0.1037)
0.0271*** (5.072)
-0.0414*** (-7.0446)
0.015*** (3.2717)
[-4, 0] 0.0079 (1.2388)
0.0306*** (5.3594)
-0.006 (0.5127)
-0.0477** (-2.2161)
0.027*** (3.2649)
0.0256*** (3.4555)
0.0118** (2.3258)
0.0323*** (5.1491)
0.0064 (0.6707)
0.0257*** (3.4292)
94
Table 6.3: Significance test between the Financial sector and each of the other nine sectors
Note: This table presents a comparison between the impact of LB bankruptcy on the Financial sector against each of the other nine sectors CAAR over the event window of [-4, 0] trading days. Sectors [-4, 0] CAAR Difference t stats P value Consumer Discretionary 0.0079 -0.0556 -3.38 0.001
Consumer Staples 0.0306 -0.0783 -4.76 <.0001 Energy -0.0060 -0.0417 -2.34 0.0211 Financials -0.0477 - - - Health Care 0.0266 -0.0743 -4.39 <.0001 Industrials 0.0256 -0.0733 -4.33 <.0001 Information Technology 0.0118 -0.0595 -3.56 0.0006
Materials 0.0323 -0.08 -4.56 <.0001 Telecommunication Services 0.0064 -0.0541 -2.61 0.0146
Utilities 0.0257 -0.0734 -3.98 0.0001
6.2 ROBUSTNESS TEST RESULTS BASED ON THE MARKET ADJUSTED
RETURN MODEL
In this section, event study results for the 10 sectors are presented based on the market
adjusted return model for an event window of [−4, 0] trading days. The market adjusted
return model is employed as an additional diagnostic check to estimate abnormal returns
for each of the 10 sectors during the event period. The statistical significance of the
AARs and CAARs were tested by employing the ordinary cross-sectional test approach
(Brown and Warner 1985). The AARs from the market adjusted return model are similar
to those obtained by employing the variant of the standard OLS MM (with an estimation
period of 150 days prior to GFC). Findings of t-test on the analysis of daily AARs and
CAARs per sector appear in Table 6.3 and 6.4 respectively. Overall, the results support
the main findings of this study. The results show that amongst the 10 sectors, the Fin
sector was the most significantly adversely affected from LB bankruptcy. There are a
few differences in terms of sign (- , +) and the level of significance of the AAR and
CAAR results (based on the market adjusted return model) presented on the following
pages compared to the AAR and CAAR results based on the variant of the standard OLS
MM (discussed earlier). Nevertheless, the slight differences result from the fact that the
OLS MM employed the Boehmer et al. (1991) standardized cross sectional test approach
95
for significance testing that is more robust than the ordinary cross sectional test approach
employed for the market adjusted return model (Boehmer et al. 1991). Since the market
adjusted return model uses the current market return to predict normal period return and
does not require an estimation period, the Boehmer et al. (1991) standardized cross
sectional test approach that requires an estimation period cannot be employed for
significance testing.
96
Table 6.4: Findings of t-tests on the analysis of daily AARs per sector based on the market adjusted return model
Notes: This table reports AARs for 10 sectors based on the market adjusted return model for an event window of [-4, 0] trading days (09/09/2008 to 15/09/2008) around the event date (15 September 2008). The S&P 500 Index is used as the market index. t - stats is reported in the parentheses. *, ** and *** represents significance at 10%, 5% and 1% levels respectively based on the ordinary cross sectional approach for significance testing. A positive sign implies positive impact from the LB bankruptcy while a negative sign implies adverse impact from the LB bankruptcy.
Days Consumer Discretionary
Consumer Staples Energy Financials Health Care Industrials Information
Technology Materials Telecommunication Services Utilities
-4 0.0041 (1.5351)
0.026*** (8.9986)
-0.0538*** (-11.3763)
-0.0204*** (-6.3582)
0.0174*** (6.5255)
-0.0035 (-0.8918)
0.0008 (0.2578)
-0.0211** (-2.6115)
-0.0284*** (-10.5981)
0.0068 (1.3861)
-3 -0.0042* (-1.9728)
-0.0064*** (-4.3845)
0.038*** (7.9199)
-0.01*** (-3.8337)
-0.0026 (-1.1324)
0.0063* (1.9822)
0.0045 (1.5458)
0.0166*** (3.3848)
-0.0399*** (-5.8827)
0.0025 (0.7769)
-2 -0.0001 (-0.0517)
-0.0024 (-0.9784)
0.005 (1.0432)
-0.0004 (-0.1535)
0.0005 (0.1734)
0.0058* (1.9853)
-0.009*** (-3.5876)
0.0024 (0.5961)
-0.0537*** (-9.6926)
-0.0063** (-2.3677)
-1 -0.0096*** (-4.3031)
0 (-0.0026)
0.0369*** (11.8284)
0.0004 (0.0811)
-0.0025 (-1.3826)
0.0078*** (3.633)
-0.0005 (-0.2227)
0.0287*** (5.4607)
-0.0607*** (-6.7995)
0.0116*** (5.0571)
0 0.0168*** (6.6472)
0.0289*** (12.3729)
-0.0448*** (-8.2367)
-0.0261*** (-3.1641)
0.0251*** (11.6766)
0.0078** (2.4307)
0.0067** (2.1712)
-0.0086 (-1.1461)
-0.0098 (-0.6883)
0.0123* (1.9222)
Table 6.5: Findings of t-tests on the analysis of CAARs per sector based on the market adjusted return model
Notes: This table reports CAARs for 10 sectors based on the market adjusted return model. The S&P 500 index is used as the market index. t - stats is reported in the parentheses. *, ** and *** represents significance at 10%, 5% and 1% levels respectively based on the ordinary cross sectional approach for significance testing. A positive sign implies positive impact from theLB bankruptcy while a negative sign implies adverse impact from the LB bankruptcy.
Window Consumer Discretionary
Consumer Staples Energy Financials Health Care Industrials Information
Technology Materials Telecommunication Services Utilities
-4 0.0041 (1.5351)
0.026*** (8.9986)
-0.0538*** (-11.3763)
-0.0204*** (-6.3582)
0.0174*** (6.5255)
-0.0035 (-0.8918)
0.0008 (0.2578)
-0.0211** (-2.6115)
-0.0068 (-1.6047)
0.0068 (1.3861)
[-4, -3] -0.0001 (-0.0183)
0.0196*** (6.7698)
-0.0158*** (-2.7498)
-0.0305*** (-7.0369)
0.0148*** (4.4184)
0.0028 (0.8327)
0.0053 (1.3753)
-0.0045 (-0.6545)
-0.0184* (-2.1372)
0.0093** (2.2701)
[-4, -2] -0.0002 (-0.0483)
0.0172*** (4.6024)
-0.0108 (-1.5768)
-0.0309*** (-6.1147)
0.0153*** (3.0758)
0.0086** (2.0449)
-0.0038 (-0.7278)
-0.0021 (-0.3071)
-0.0322*** (-4.0629)
0.0029 (0.5659)
[-4, -1] -0.0098** (-2.4847)
0.0172*** (4.1063)
0.0262*** (3.5141)
-0.0305*** (-3.7168)
0.0128** (2.5081)
0.0164*** (3.4173)
-0.0042 (-0.7912)
0.0266*** (4.0894)
-0.0391** (-3.571)
0.0145*** (3.06)
[-4, 0] 0.007 (1.4251)
0.0461*** (8.9857)
-0.0186** (-2.2618)
-0.0566*** (-3.6536)
0.038*** (7.1679)
0.0242*** (3.8952)
0.0025 (0.3967)
0.0181* (1.8334)
0.0117 (0.7158)
0.0269** (2.7293)
97
6.3 CONCLUSION
The event study results generally support the first proposition that the financial sector
was the most significantly adversely affected from the LB bankruptcy. Amongst the 10
sectors, the Fin sector is not only found to be significantly adversely affected on the day
of LB bankruptcy but also during the multi day interval of [−4, 0] trading days. Evidence
in support of sector heterogeneity and market efficiency is also found. The market
adjusted return model that was employed as a robustness test confirms the main findings
of this study and supports the first proposition.
98
CHAPTER 7: EVENT STUDY RESULTS FOR THE FINANCIAL INDUSTRIES
7.0 INTRODUCTION
In this chapter, the event study results for the 4 financial industries are presented. The
abnormal return performances of all the 4 financial industries were analyzed to
determine whether the diversified financial industry was the most significantly adversely
affected during the LB bankruptcy. The event study results are based on the variant of
the standard OLS MM with an estimation period before 30 June 2007 and the robustness
test results are based on the market adjusted return model. For each of the 4 financial
industries, the daily AAR is reported for a 5 day event window (from −4 to 0 trading
days46 ). In the table, t - stats is reported in the parentheses and *, ** and *** represents
significance at 10%, 5% and 1% levels respectively. The Boehmer et al. (1991)
standardized cross sectional test approach is used for significance testing. A positive
sign implies positive impact from the LB bankruptcy while a negative sign implies
adverse impact from the LB bankruptcy. Furthermore, the evolution of CAAR results
over the multi day intervals of [−4, −3], [−4, −2], [−4, −1] and [−4, 0] are presented.
This chapter has three main sections. In Section 7.1, AAR and CAAR results for the 4
financial industries, which are based on the OLS MM and an estimation period of 150
days prior to 30 June 2007, are discussed. Section 7.2 presents the robustness test results
based on the market adjusted return model. Finally, the chapter conclusion is provided in
Section 7.3.
46 9 September 2008 to 15 September 2008
99
7.1 THE OLS MM RESULTS BASED ON AN ESTIMATION PERIOD OF 150
DAYS BEFORE 30 JUNE 2007
In this section, event study results for the financial industries are presented. The event
study is based on the OLS MM with an estimation period of 150 days prior to 30 June
2007 (22 November 2006 to 29 June 2007) and an event window of [−4, 0] trading days
(9 September 2008 to 15 September 2008). Findings of t-test on the analysis of daily
AARs and CAARs per industry appear in Tables 7.1 and 7.2 respectively. Overall, the
AAR and CAAR results show that the significant adverse impact was discriminatory
towards the diversified financial industry, which was the most exposed to LB. This
provides evidence in support of my proposition that the diversified financial industry’s
performance was the most significantly adversely affected during the LB bankruptcy.
The main findings are summarized as follows. Further to the study by Narayan and
Sharma (2011), this study provides evidence on industry heterogeneity during the LB
bankruptcy. Although the performance of all financial industries was adversely affected
on the event date (depicted by negative AAR) and over the window of [−4, 0] trading
days, not all financial industries’ performance were significantly adversely affected. On
the event date, only diversified financials and REIT performances were significantly
adversely affected. The diversified financials AAR (−1.70%) and REIT AAR (−1.88%)
are significantly affected at the 1% level of significance. Additionally, the CAAR results
for the [−4, 0] trading days window show that only diversified financials was
significantly adversely affected at the 1% level. The other three financial industries were
insignificantly affected during the [−4, 0] trading days window. This implies that
although all financial industries were adversely affected during the LB bankruptcy, the
adverse impact was not significant on all financial industries. In support, the findings of
Dumontaux and Pop (2009) ‘‘suggest that the observed contagious effects were
rational/discriminating rather than panic-driven/undifferentiated and tend to weaken the
case for the bailout of Lehman Brothers’’.
100
Additionally, the AAR and CAAR results show that the diversified financials industry’s
performance correlates directly with negative news announcements on LB. The trading
days on which negative information were released included 9 September 2008, 10
September 2008, 12 September 2008 and 15 September 2008 (Bartram and Bodnar
2009). These are −4, −3, −1 and 0 trading days respectively. The daily AAR shows that
the diversified financial industry is the only financial industry that suffered consistent
adverse effects on −4, −3, −1 and 0 trading days. This results in diversified financial
industry having the most number of significant negative AARs from −4 trading days till
the event date. In support, the CAAR results show that the diversified financials
industry’s performance was the most significantly adversely affected over the [−4, 0]
trading days event window. This also provides evidence in support of a semi-strong
form of market efficiency by showing a direct relationship between the negative news
announcement on LB and the price return performances of the diversified financial
industry. Diversified financials is the only industry having consistent negative CAAR
significant at the 1% level over the multi day intervals of [−4, −3], [−4, −2], [−4, −1] and
[−4, 0] trading days. This implies that since LB was an investment bank, negative news
announcements on LB greatly affected the confidence among the diversified financial
industry’s investors in comparison to other financial industries. In support, based on the
market adjusted return model, Dumontaux and Pop (2009) find the components of
disaggregated diversified financial performance (investment services and diversified
financial services firm) to be the most significantly adversely affected by the LB
bankruptcy. Therefore, this study’s findings using the variant of the standard Fama et al.
(1969) market model and a larger sample representation47 complement those of the
study by Dumontaux and Pop (2009).
47 This study examines all four financial industries, namely, banks, diversified financials, REIT and Insurances, from NYSE. On the other hand, Dumontaux and Pop (2009) examined only large disaggregated bank and non-bank financial institutions.
101
Table 7.1: Findings of t-tests on the analysis of daily AARs per financial industry based on the variant of the standard OLS MM
Note: This table reports AARs for financial industries based on the OLS MM with 150 days estimation period prior to 30 June 2007 (22/11/2006 to 29/06/2007) and event window of [-4, 0] trading days (9/9/2008 to 15/9/2008). t - stats is reported in the parentheses. *, ** and *** represents significance at 10%, 5% and 1% levels respectively based on the Boehmer et al. (1991) standardized cross sectional approach for significance testing. A positive sign implies positive impact from the LB bankruptcy while a negative sign implies adverse impact from the LB bankruptcy.
Days Banks Diversified Financials Insurance REIT
-4 -0.0184*** (-4.2772)
-0.0253*** (-4.7448)
-0.0206* (-2.049)
0.0014 (0.2307)
-3 -0.0241***
(-4.396) -0.0076** (-2.154)
-0.0017 (-0.6727)
-0.002 (-0.424)
-2 0.0129** (2.6625)
-0.0072 (-1.504)
-0.0088** (-2.2172)
-0.002 (-0.7461)
-1 0.0234*** (3.1293)
-0.0096* (-1.9551)
-0.0208 (-1.2545)
0.0129*** (3.5925)
0 -0.016
(-1.4368) -0.017** (-2.1117)
-0.0303 (-1.04)
-0.0188** (-2.4373)
Table 7.2: Findings of t-tests on the analysis of CAARs per financial industry based on the variant of the standard OLS MM
Notes: This table reports CAARs for financial industries based on the OLS MM with 150 days estimation period prior to 30 June 2007 (22/11/2006 to 29/06/2007) for various multi day intervals leading to the event date (15 September 2008). t - stats is reported in the parentheses. *, ** and *** represents significance at 10%, 5% and 1% levels respectively based on the Boehmer et al. (1991) standardized cross sectional approach for significance testing. A positive sign implies positive impact from the LB bankruptcy while a negative sign implies adverse impact from the LB bankruptcy.
Windows Banks Diversified Financials Insurance REIT
-4 -0.0184*** (-4.2772)
-0.0253*** (-4.7448)
-0.0206* (-2.049)
0.0014 (0.2307)
[-4, -3] -0.0425*** (-5.4218)
-0.0328*** (-6.1915)
-0.0223 (-1.7016)
-0.0006 (-0.3204)
[-4, -2] -0.0296** (-2.5349)
-0.0401*** (-4.6)
-0.0311** (-2.1002)
-0.0026 (-0.6831)
[-4, -1] -0.0062 (-0.4524)
-0.0497*** (-4.5309)
-0.0519 (-1.6551)
0.0104 (1.6529)
[-4, 0] -0.0222 (-1.2395)
-0.0668*** (-4.4699)
-0.0822 (-1.3574)
-0.0085 (-0.6146)
102
7.2 ROBUSTNESS TEST RESULTS BASED ON THE MARKET ADJUSTED
RETURN MODEL
In this section, event study results for the 4 financial industries are presented based on
the market adjusted return model for an event window of [−4, 0] trading days. The
market adjusted return model is employed as an additional diagnostic check to estimate
abnormal returns for each of the 4 financial industries during the event period. The
statistical significance of the AARs and CAARs were tested by employing the ordinary
cross-sectional test approach (Brown and Warner 1985). The AARs from the market
adjusted return model are similar to those obtained by employing the OLS MM with an
estimation period of 150 days prior to GFC. Findings of t-test on the analysis of daily
AARs and CAARs per financial industry appear in Table 7.3 and 7.4 respectively.
Overall, the results support the main findings of this study. The results show that
amongst the 4 financial industries, the diversified financial industry was the most
significantly adversely affected from LB bankruptcy. There are a few differences in
terms of sign (- , +) and the level of significance of the AAR and CAAR results (based
on the market adjusted return model) presented on the following pages compared to the
AAR and CAAR results based on the variant of the standard OLS MM (discussed in
Section 7.1). Nevertheless, the slight differences result from the fact that the OLS MM
employed the standardized cross sectional test approach for significance testing that is
more robust than the ordinary cross sectional test approach employed for the market
adjusted return model (Boehmer et al. 1991). Since the market adjusted return model
uses the current market return to predict normal period return and does not require an
estimation period, the standardized cross sectional test approach that requires an
estimation period cannot be employed for significance testing.
103
Table 7.3: Findings of t-tests on the analysis of daily AARs per financial industry based on the market adjusted return model
Note: This table reports AARs for financial industries based on the market adjusted return model and event window of [-4, 0] trading days (9/9/2008 to 15/9/2008) around the event date (15 September 2008). t - stats is reported in the parentheses. *, ** and *** represents significance at 10%, 5% and 1% levels respectively based on the ordinary cross sectional approach for significance testing. A positive sign implies positive impact from the LB bankruptcy while a negative sign implies adverse impact from the LB bankruptcy.
Days Banks Diversified Financials Insurance REIT
-4 -0.0173*** (-3.8607)
-0.0377*** (-6.0487)
-0.0155* (-1.8221)
-0.0102** (-2.9257)
-3 -0.0252***
(-4.597) -0.0054
(-1.1236) -0.0024
(-0.6678) -0.001
(-0.1835)
-2 0.0114** (2.1801)
-0.0022 (-0.4722)
-0.0107** (-2.7881)
0.0013 (0.3507)
-1 0.0225***
(3.127) -0.0089
(-1.6484) -0.0209
(-1.3349) 0.0126*** (4.1575)
0 -0.0142 (-1.3)
-0.0341*** (-3.922)
-0.0233 (-0.8074)
-0.0345*** (-3.6904)
Table 7.4: Findings of t-tests on the analysis of CAARs per financial industry based on the market adjusted return model
Note: This table reports CAARs for Financial industries based on the market adjusted return model for various multi day intervals leading to the event date (15 September 2008). t - stats is reported in the parentheses. *, ** and *** represents significance levels at 10%, 5% and 1% level respectively based on the ordinary cross sectional approach for significance testing. A positive sign implies positive impact from the LB bankruptcy while a negative sign implies adverse impact from the LB bankruptcy.
Windows Banks Diversified Financials Insurance REIT
-4 -0.0173*** (-3.8607)
-0.0377*** (-6.0487)
-0.0155* (-1.8221)
-0.0102** (-2.9257)
[-4, -3] -0.0425*** (-5.2547)
-0.043*** (-7.2063)
-0.0179 (-1.5361)
-0.0113* (-1.8344)
[-4, -2] -0.0312*** (-2.9679)
-0.0453*** (-5.3386)
-0.0285** (-2.2451)
-0.01 (-1.4658)
[-4, -1] -0.0087 (-0.8981)
-0.0542*** (-5.4312)
-0.0495* (-1.8377)
0.0026 (0.3519)
[-4, 0] -0.0229 (-1.3404)
-0.0883*** (-5.5483)
-0.0728 (-1.3169)
-0.0318** (-2.3649)
104
7.3 CONCLUSION
The event study results generally support the second proposition that the diversified
financial industry’s performance was the most significantly adversely affected from the
LB bankruptcy. The diversified financial industry’s performance was not only
significantly adversely affected on the event date but also it was the most significantly
adversely affected throughout the multi day intervals of [−4, −3], [−4, −2], [−4, −1] and
[−4, 0] trading days. The results also provide evidence in support of industry
heterogeneity. Although all financial industries’ performance were adversely affected on
the event date (depicted by negative AAR) and during the [−4, 0] multi-day interval, not
all financial industries’ performance were significantly adversely affected. The
significant adverse impact from the LB bankruptcy was discriminatory towards the
industry most exposed to LB. Finally, this study finds evidence in support of a semi
strong form of market efficiency by showing a direct relationship between the negative
news announcement on LB and the price return performances of the diversified financial
industry. The market adjusted return model that was employed as a robustness test
confirms the main findings of this study and supports the second proposition.
105
CHAPTER 8: CONCLUSION
Implications and Future Research
8.0 INTRODUCTION
The 150 years old LB was the fourth largest investment bank in the US. It was
considered one of the Wall Street's biggest dealers in fixed-interest trading and had
heavily invested in securities linked to the US sub-prime mortgage market. The
catastrophic collapse of LB (with $639 billion worth of total assets) was the biggest
failure in the US history (Dumontaux and Pop 2009) and marks the anticlimax era in the
performance of stock markets globally. LB bankruptcy triggered incredible declines in
index levels and escalated price volatility universally (Bartram and Bodnar 2009; Chong
2011; Eichler et al. 2011; Samarakoon 2011).
This study extends the work of Bartram and Bodnar (2009) with the use of statistical
methods (hypothesis testing, event study approach and significance tests) and thoroughly
examines the impact of the LB bankruptcy on the performance of NYSE sectors and
financial industries. In light of the idea of firm and sector heterogeneity (Narayan and
Sharma 2011), this study investigates whether the financial sector and the Diversified
financial industry were the most significantly adversely affected from the LB
bankruptcy. This study employs the following event study approaches to investigate the
abnormal return performance of each of the 10 sectors and 4 financial industries. (1) The
standard OLS MM with an estimation period of 150 days immediately before the event
window of [−30, +30] trading days. (2) A variant of the standard OLS MM with an
estimation period of 150 days before the GFC (30 June 2007). Robustness tests for the
event study results are also conducted with the use of market adjusted return model.
Standardized cross sectional approach is used for testing the significance of the OLS
MM results while ordinary cross sectional approach is employed to test the significance
of the results based on the market adjusted return model.
106
This chapter concludes this study by highlighting the main findings, research limitations
and implications for future research. The remainder of this chapter is organized as
follows. In Section 8.1, the main findings of this study are discussed. Next, the
implications of this study are discussed in Section 8.2. Finally, the limitations of this
research and suggestions for future research are provided in Section 8.3.
8.1 MAIN FINDINGS
The event study results generally support both the proposition and show that the
performance of the financial sector and the diversified financial industry were the most
significantly adversely affected from the LB bankruptcy. The main contribution of this
study is that it provides evidence that sectors behave heterogeneously during a crisis and
unravels that the significant adverse impact from the LB bankruptcy is discriminatory
towards the financial sector and the diversified financial industry that were most
exposed to LB.
This study unravels two new findings during LB bankruptcy. First, this study provides
evidence that the impact from the LB bankruptcy varied across sectors and financial
industries. On the event date and over the [−4, 0] trading days window, being the period
over which most negative news on LB were released, not all sectors and financial
industries’ performance were significantly adversely affected. Although the performance
of all financial industries were adversely affected on the event date and during the multi
day interval of [−4, 0] trading days, not all financial industries’ performance were
significantly adversely affected. Second, this study finds negative news on the LB
bankruptcy having the greatest adverse impact on the performance of the financial sector
and the diversified financial industry (both were most exposed to LB) in comparison to
other sectors and financial industries. The financial sector and the diversified financial
industry’s performance were significantly adversely affected not only on the event date
but also the [−4, 0] trading days event window.
107
Additionally, this study finds evidence in support of a semi strong form of market
efficiency. Since the financial sector and the diversified financial industry were the most
significantly correlated with LB, negative news announcements on LB had a significant
adverse impact on the performance of the financial sector and the diversified financial
industry. These findings imply that the US share market and investors instantly reacted
to the negative news announcements on LB. In support, Pichardo and Bacon (2009) with
the use of 15 investment firms also find the US stock market semi strong form
informational efficient during the LB bankruptcy.
The event study results also reveal that the standard OLS MM provides biased estimates
in predicting normal period returns for the financial sector. Consequently, the biased
estimation parameters lead to inaccurate positive abnormal return performance for the
financial sector. On the other hand, the use of a variant of the standard OLS MM with an
estimation period of 150 days before the GFC (before 30 June 2007) provides better
results.
8.2 IMPLICATIONS
The findings of this study have economic implications for investors and policy makers.
The economic implications for investors are discussed as follows. The findings of this
study suggest that sectors behave heterogeneously during a crisis and significant adverse
impact from the LB bankruptcy is discriminatory towards the financial sector and the
diversified financial industry that were the most exposed to LB. Therefore, investors
could utilize these findings to devise profitable trading strategies.
Furthermore, this study’s findings imply that the impact of a financial shock arising
from a particular company would be most significant on those industries and sectors
which have been more exposed to the shock company. In other words, the greater the
exposure of the industries and sectors to the shock company, the more volatility the
108
industries and sectors will be exposed to. For instance, this study provides evidence that
the impact of LB bankruptcy was most significant on the financial sector and more
specifically the diversified financial industry because they were the most exposed to LB.
The transmission effect during the 2000 Nasdaq technology bubble collapse was TMT
sector specific and not widespread on other sectors (Hon et al. 2007). The 2000 Nasdaq
technology bubble collapse also reveals that the TMT sector was most affected because
it was most exposed to the collapse (Hon et al. 2007). Therefore, future investors should
always consider diversifying their investment portfolios across sectors and industries
rather than investing exclusively in any particular sector or industry (investors should
not be sector or industry specific). During LB bankruptcy, those investors who had
solely invested in the diversified financials industry or the financial sector would have
suffered more drastically in comparison to those who had diversified their investment
across industries and sectors. Therefore, if a shock takes place in a particular company,
diversification benefits at industry and sector level could still be achieved.
Additionally, this study shows that for the firms in the financial sector, the standard OLS
MM predicted much lower returns than what the actual return turned out to be. As a
result, the financial sector suffered positive abnormal returns during the high volatile
period of LB bankruptcy. Conversely, the variant of the standard OLS MM, which has
an estimation period of 150 days before GFC (before 30 June 2007), provides better
results. This implies that when the event period as well the periods surrounding the event
window are characterized by high levels of volatility, it will be more appropriate to
choose an estimation period immediately before the volatile period (e.g. as used in this
study based on the variant of the standard OLS MM) as opposed to choosing an
estimation period immediately before the event period (as typically used in the standard
OLS MM).
Moreover, the economic implication for the policymakers is as follows. Hon et al.
(2007) find that the impact of the 2000 Nasdaq technology bubble collapse in the US
109
TMT sector was limited to TMT sectors with sector specific contagion effects in non US
markets. This study finds that the adverse effect from LB bankruptcy, which occurred in
the financial sector, was most significant on the financial sector itself. Amongst the
financial industries, the diversified financial industry was most significantly adversely
affected from the LB bankruptcy. In other words, LB most significantly affected the
diversified financials at the financial industry level. Next, LB bankruptcy most
significantly affected the financial sector and via contagion effects global stock market
performance was adversely affected. This holds strong economic implications towards
the need for identifying the dominant industries and sectors during a crisis and
accordingly policymakers should design policies to protect other non influential sectors
and industries.
In times of a crisis a “special crisis committee” could be formed by the government in
collaboration with relevant authorities to oversee and investigate upon the potential
problematic firms within the dominant sectors. Specifically, the special crisis committee
could look into issues such as ethics, corporate governance, accounting and management
in order to suggest possible strategies to strengthen the potential problematic firms. As a
result, corporate collapses could be avoided or minimized without the need of bailouts.
Accordingly, tax payers’ money would be saved for better use in terms of economic
growth and development. It is also noteworthy that the special crisis committee should
have a wider representation in terms of relevant authorities to bring a wide range of
views, ideas, expertise and information about the potential problematic firms within the
dominant sectors. This will not only enhance the devising of sound strategies but also
avoid domination by a certain group of self interested authorities. Furthermore,
companies exposed to the shock company can avoid adverse contagion effects by
diversifying their investments. This will promote sustainable business growth and
minimize the risks (potential systematic risks) associated with unmanageable business
growth (businesses becoming “too big” and failing). In other words, if investors
diversify their investment across firms, industries and sectors, these economic units will
also grow at a sustainable rate. On the other hand, if investors heavily invest in a
110
particular firm (e.g. a bank) then the firm will become “too big” and unmanageable. This
will be detrimental to the economy at large as evidenced by the recent GFC. Therefore,
policymakers should advocate on the essence of holding diversified investments.
8.3 LIMITATIONS AND SUGGESTIONS FOR FUTURE RESEARCH
This study used an event study approach to analyze the performance of the sectors and
financial industries during LB bankruptcy. Under this approach the abnormal
performance of firms during the event period is only attributed to the event under
investigation. As a result, various other factors or events that may have influenced firm
returns during the event period other than the events under investigation are not
considered explicitly. This is not only the limitation of this study but generally the
limitation of the event study paradigm. Nevertheless, in event studies, it is a common
practice to gather a sufficiently large sample of firms experiencing the event to ensure
that the single commonality among the firms is the event. Consequently, other random
factors are canceled and the statistical tests appropriately reflect the impact of the event
under investigation. Also, in this study, all necessary measures were taken to ascertain
reliable results under conditions of event-induced variances. Firstly, the most popular
Fama et al. (1969) MM was used to do the analysis. Secondly, the robust hybrid
Boehmer et al. (1991) standardized cross sectional test was conducted to ensure that the
hypothesis testing is done correctly. Prior studies advocate the use of hybrid
standardized cross sectional test (more robust) as being well specified for hypothesis
testing under conditions of event induced variances (Boehmer et al. 1991). Finally, the
market adjusted return model was employed to check the robustness of the event study
results
Finally, this study finds that sector heterogeneity existed during LB bankruptcy and has
broad implications for investors in terms of using the findings of this study to devise
profitable trading strategy and holding diversified investments across sectors. Future
111
research could analyze and provide suggestions for investors on, for instance, the best
investor strategy to yield maximum benefits during a crisis.
REFERENCE
Acharya, V., T. Philippon, M. Richardson, and N. Roubini. 2009. The Financial Crisis of 2007 - 2009: Causes and Remedies. Financial Markets, Institutions & Instruments 18 (2):89-137.
Aggarwal, R., C. Inclan, and R. Leal. 1999. Volatility in emerging stock markets. Journal of Financial and Quantitative Analysis 34 (01):33-55.
Ahlgren, N., and J. Antell. 2010. Stock market linkages and financial contagion: A cobreaking analysis. The Quarterly Review of Economics and Finance 50 (2):157-166.
Ammann, M., and S. Kessler. 2008. Intra-Day Characteristics of Stock Price Crashes. Applied Financial Economics 19 (15):1239 - 1255.
Angkinand, A. P., J. R. Barth, and H. Kim. 2009. Forthcoming in the Financial and Economic Crises: An International Perspective.
Arshanapalli, B., J. Doukas, and L. H. P. Lang. 1995. Pre and post-October 1987 stock market linkages between US and Asian markets. Pacific-Basin Finance Journal 3 (1):57-73.
Baig, T., and I. Goldfajn. 1999. Financial market contagion in the Asian crisis. IMF staff papers:167-195.
Baillie, R., and P. DeGennaro. 1990. Stock return and volatility. Journal of Financial and Quantitative Analysis 25 (2):203 – 214.
Bakshi, G., C. Cao, and Z. Chen. 1997. Empirical performance of alternative option pricing models. Journal of finance:2003-2049.
Barber, B. M., and J. D. Lyon. 1997. Detecting long-run abnormal stock returns: The empirical power and specification of test statistics. Journal of Financial Economics 43 (3):341-372.
Bartholomeusz, S. 2008. Why Lehman was allowed to fail. Business Spectator. Bartram, S. M., and G. M. Bodnar. 2009. No place to hide: The global crisis in equity
markets in 2008/2009. Journal of International Money and Finance 28 (8):1246-1292.
Basher, S. A., M. K. Hassan, and A. M. Islam. 2007. Time-varying volatility and equity returns in Bangladesh stock market. Applied Financial Economics 17 (17):1393-1407.
Bates, D. S. 1997. Post-'87 crash fears in S&P 500 futures options: National Bureau of Economic Research Cambridge, Mass., USA.
Bautista, C. C. 2005. How volatile are East Asian stocks during high volatility periods? Applied Economics Letters 12 (5):319-326.
Beal, D., and W. McKeown. 2009. Personal Finance. 4 ed. 42 McDougall Street, Milton Qld 4064: Wiley & Sons Australia, Ltd.
Beaver, W. H. 1968. The information content of annual earnings announcements. Journal of accounting research 6 (3):67-92.
Bekaert, G., and G. Wu. 2000. Asymmetric volatility and risk in equity markets. Review of Financial Studies 13 (1):1-42.
112
Black, F. 1976. Stuedies of stock price volatility changes. Proceedings of the 1976 Meetings of the American Statistical Association, Business and Economical Statistics Section:177 - 181.
Boehmer, E., J. Masumeci, and A. B. Poulsen. 1991. Event-study methodology under conditions of event-induced variance. Journal of Financial Economics 30 (2):253-272.
Boudoukh, J., M. Richardson, and R. F. Whitelaw. 1997. Nonlinearities in the relation between the equity risk premium and the term structure. Management Science 43 (3):371-385.
Breen, W., L. R. Glosten, and R. Jagannathan. 1989. Economic significance of predictable variations in stock index returns. Journal of finance 44 (5):1177-1189.
Brisbois, F., P. Boveroux, M. Ausloos, and N. Vandewalle. 2000. On the Occurrence of Financial Crashes. International Journal of Theoretical and Applied Finance 3 (3):423-424.
Brooks, C. 2014. Introductory econometrics for finance: Cambridge university press. Brown, S. J., and J. B. Warner. 1980. Measuring security price performance. Journal of
Financial Economics 8 (3):205-258. Brown, S. J., and J. B. Warner. 1985. Using daily stock returns: The case of event
studies. Journal of Financial Economics 14 (1):3-31. Cable, J., and K. Holland. 1996. Modelling Normal Returns in Event Studies: A Model-
Selection Approach and Pilot Study. ABERYSTWYTH ECONOMIC RESEARCH PAPERS.
Cable, J., and K. Holland. 1999. Regression vs. non – regressive models of normal returns: implications for event studies. Economics Letters 64 (1):81 – 85.
Calice, G. 2011. The subprime asset-backed securities market and the equity prices of large complex financial institutions. Journal of International Financial Markets, Institutions and Money 21 (4):585 - 604.
Campbell, J. Y. 1987. Stock returns and the term structure. Journal of Financial Economics 18 (2):373-399.
Campbell, J. Y., and L. Hentschel. 1992. No news is good news:: An asymmetric model of changing volatility in stock returns. Journal of Financial Economics 31 (3):281-318.
Caporale, G. M., N. Pittis, and N. Spagnolo. 2006. Volatility transmission and financial crises. Journal of Economics and Finance 30 (3):376-390.
Celikkol, H., S. Akkoc, and Y. D. Akarim. 2010. The Impact of Bankruptcy of Lehman Brothers on the Volatility Structure of ISE- 100 Price Index. Journal of Money, Investment and Banking (18):5-12.
Chakrabarty, B., and G. Zhang. 2010. Financial contagion channels: Market microstructure evidence from Lehman Brothers Bankruptcy, 1-45.
Chen, C. W. S., T. C. Chiang, and M. K. P. So. 2003. Asymmetrical reaction to US stock-return news: evidence from major stock markets based on a double-threshold model. Journal of Economics and Business 55 (5-6):487-502.
Chen, C. W. S., M. J. Yang, R. Gerlach, and H. Jim Lo. 2006. The asymmetric reactions of mean and volatility of stock returns to domestic and international information based on a four-regime double-threshold GARCH model. Physica A: Statistical Mechanics and its Applications 366:401-418.
113
Cheung, W., S. Fung, and S. C. Tsai. 2010. Global capital market interdependence and spillover effect of credit risk: evidence from the 2007-2009 global financial crisis. Applied Financial Economics, 20 1 (2):85-103.
Chong, C. Y. 2011. Effect of Subprime Crisis on US Stock Market Return and Volatility. Global Economy and Finance Journal 4 (1):102-111
Chou, R. Y. 1988. Volatility persistence and stock valuations: Some empirical evidence using GARCH. Journal of Applied Econometrics 3 (4):279-294.
Choudhry, T. 1996. Stock market volatility and the crash of 1987: evidence from six emerging markets. Journal of International Money and Finance 15 (6):969-981.
———. 2000. Meltdown of 1987 and meteor showers among Pacific-Basin stock markets. Applied Financial Economics 10 (1):71-80.
Christie, A. A. 1982. The stochastic behavior of common stock variances:: Value, leverage and interest rate effects. Journal of Financial Economics 10 (4):407-432.
Coelho, L., and R. J. Taffler. 2008. When is bad news really good news? The case of strategic vs. non-strategic bankruptcies, 1-50.
Cotter, J. 2006. Extreme value estimation of boom and crash statistics. European Journal of Finance 12 (6-7):553-566.
Cuñado, J., J. Gomez Biscarri, and F. Perez de Gracia. 2006. Changes in the dynamic behavior of emerging market volatility: Revisiting the effects of financial liberalization. Emerging Markets Review 7 (3):261-278.
De la Torre, A., J. C. Gozzi, and S. L. Schmukler. 2007. Stock market development under globalization: Whither the gains from reforms? Journal of Banking & Finance 31 (6):1731-1754.
De Santis, G., and S. Imrohoroglu. 1997. Stock returns and volatility in emerging financial markets. Journal of International Money and Finance 16 (4):561-579.
Dissanaike, G., and A. Le Fur. 2003. On the use of the log car measure in event studies. Journal of Business Finance & Accounting 30 (7 8):1165-1170.
Dufrenot, G., V. Mignon, and A. Peguin-Feissolle. 2011. The Effects of the Subprime Crisis on the Latin American Financial Markets: An Empirical Assessment, 1-40.
Dumontaux, N., and A. Pop. 2009. Contagion Effects in the Aftermath of Lehman's Collapse: Measuring the Collateral Damage: mimeo, 1-26.
Ederington, L. H., and W. Guan. 2010. How asymmetric is US stock market volatility? Journal of Financial Markets 13 (2):225-248.
Edwards, S., J. G. Biscarri, and F. P. Gracia. 2003. Stock market cycles, financial liberalization and volatility. Journal of International Money and Finance 22 (7):925 - 955.
Eichler, S., A. Karmann, and D. Maltritz. 2011. The term structure of banking crisis risk in the United States: A market data based compound option approach. Journal of Banking & Finance 35 (4):876-885.
Fama, E., L. Fisher, M. Jensen, and R. Roll. 1969. The adjustment of stock prices to new information. International Economic Review 10 (1):1 - 21.
Fama, E. F., and K. R. French. 1989. Business conditions and expected returns on stocks and bonds. Journal of Financial Economics 25 (1):23-49.
———. 2004. The capital asset pricing model: theory and evidence. The Journal of Economic Perspectives 18 (3):25-46.
114
Fama, E. F., and G. W. Schwert. 1977. Asset returns and inflation. Journal of Financial Economics 5 (2):115-146.
Fang, W. 2001. Stock return process and expected depreciation over the Asian financial crisis. Applied Economics 33 (7):905-912.
Foster, G. 1980. Accounting policy decisions and capital market research. Journal of Accounting and Economics 2 (1):29-62.
French, K. R., G. W. Schwert, and R. F. Stambaugh. 1987. Expected stock returns and volatility. Journal of Financial Economics 19 (1):3 - 29.
Gklezakou, T., and J. Mylonakis. 2009. Interdependence of the developing stock markets, before and during the economic crisis: The case of south Europe. Journal of Money, Investment and Banking 11 (4):70-78.
Glosten, L. R., R. Jagannathan, and D. E. Runkle. 1993. On the relation between the expected value and the volatility of the nominal excess return on stocks. Journal of finance:1779-1801.
Gorban, A. N., E. V. Smirnova, and T. A. Tyukina. 2010. Correlations, risk and crisis: From physiology to finance. Physica A: Statistical Mechanics and its Applications 389 (16):3193-3217.
Grossman, S. J. 1988. An analysis of the implications for stock and futures price volatility of program trading and dynamic hedging strategies. Journal of business 61 (3):275 - 298.
Guo, F., C. R. Chen, and Y. S. Huang. 2011. Markets contagion during financial crisis: A regime-switching approach. International Review of Economics & Finance 20 (1):95-109.
Hale, D. D. 1994. Stock markets in the new world order. The Columbia Journal of World Business 29 (2):14-28.
Hammoudeh, S., and H. Li. 2008. Sudden changes in volatility in emerging markets: The case of Gulf Arab stock markets. International Review of Financial Analysis 17 (1):47-63.
Head, T. 2008. CAPM: Theory, Advantages and Dis advantages. The Journal of Finance 27 (5):175-197.
Hon, M. T., J. K. Strauss, and S. K. Yong. 2007. Deconstructing the Nasdaq bubble: A look at contagion across international stock markets. Journal of International Financial Markets, Institutions and Money 17 (3):213-230.
Hsin, C. W. 2004. A multilateral approach to examining the comovements among major world equity markets. International Review of Financial Analysis 13 (4):433-462.
Huang, B. N., and C. W. Yang. 2000. The Impact of Financial Liberalization on Stock Price Volatility in Emerging Markets* 1. Journal of Comparative Economics 28 (2):321-339.
Iwata, S., and S. Wu. 2009. Stock market liberalization and international risk sharing. International Financial Markets, Institutions and Money 19 (3):461-476.
James, G. A., and M. Karoglou. 2010. Financial liberalization and stock market volatility: the case of Indonesia. Applied Financial Economics 20 (6):477-486.
Kassimatis, K. 2002. Financial liberalization and stock market volatility in selected developing countries. Applied Financial Economics 12 (6):389-394.
Kaufman, G. G. 1994. Bank contagion: A review of the theory and evidence. Journal of Financial Services Research 8 (2):123-150.
115
Kawakatsu, H., and M. R. Morey. 1999. An empirical examination of financial liberalization and the efficiency of emerging market stock prices. Journal of Financial Research 22 (4):385-411.
Kearney, C. 2000. The determination and international transmission of stock market volatility. Global Finance Journal 11 (1-2):31-52.
Kenourgios, D., A. Samitas, and N. Paltalidis. 2011. Financial crises and stock market contagion in a multivariate time-varying asymmetric framework. Journal of International Financial Markets, Institutions and Money 21 (1):92–106.
Kim, E. H., and V. Singal. 2000. The fear of globalization markets. Emerging Markets Review 1 (3):183 - 198.
King, M. R. 2001. Who triggered the Asian financial crisis? Review of International Political Economy 8 (3):438-466.
Kothari, S., and J. B. Warner. 1997. Measuring long-horizon security price performance. Journal of Financial Economics 43 (3):301-339.
Kryzanowski, L., L. Switzer, and L. Jiang. 1995. Stock market crash behavior of screen-sorted portfolios. International Review of Economics & Finance 4 (3):227-244.
Kwan, F. B., and M. G. Reyes. 1997. Price effects of stock market liberalization in Taiwan. The Quarterly Review of Economics and Finance 37 (2):511-522.
Law, S. H. 2006. Has Stock Market Volatility in the Kuala Lumpur Stock Exchange Returned to Pre-Asian Financial Crisis Levels? ASEAN Economic Bulletin 23 (2):212-229.
Lee, C. F., G. Chen, and O. M. Rui. 2001. Stock returns and volatility on China's stock markets. Journal of Financial Research 24 (4):523-544.
Lee, S., and D. J. Connolly. 2010. The impact of IT news on hospitality firm value using cumulative abnormal returns (CARs). International Journal of Hospitality Management 29 (3):354-362.
Lintner, J. 1965. The valuation of risk assets and the selection of risky investments in stock portfolios and capital budgets. The review of economics and statistics:13-37.
Liu, Y. A., M. S. Pan, and J. C. P. Shieh. 1998. International transmission of stock price movements: Evidence from the US and five Asian-Pacific markets. Journal of Economics and Finance 22 (1):59-69.
Longstaff, F. A. 2010. The subprime credit crisis and contagion in financial markets. Journal of Financial Economics 97 (3):436-450.
MacKinlay, A. C. 1997. Event studies in economics and finance. Journal of economic literature 35 (1):13-39.
Mio, C., and M. Fasan. 2012. Does Corporate Social Performance Yield Any Tangible Financial Benefit During a Crisis? An Event Study of Lehman Brothers' Bankruptcy. Corporate Reputation Review 15 (4):263-284.
Nam, J. H., K. Yuhn, and S. B. Kim. 2008. What happened to pacific-basin emerging markets after the 1997 financial crisis? Applied Financial Economics 18 (8):639-658.
Narayan, P. K., and S. S. Sharma. 2011. New evidence on oil price and firm returns. Journal of Banking & Finance 35 (12):3253-3262.
Nelson, D. B. 1991. Conditional heteroskedasticity in asset returns: a new approach. Econometrica: Journal of the Econometric Society 59 (2):347-370.
116
Nikkinen, J., M. M. Omran, P. Sahlstrom, and J. Aijo. 2008. Stock returns and volatility following the September 11 attacks: Evidence from 53 equity markets. International Review of Financial Analysis 17 (1):27-46.
Pagan, A. R., and Y. S. Hong. 1991. Nonparametric estimation and the risk premium in William Barnett, James Powell and George Tauchan, eds: Non parametric and semiparametric Methods in Econometrics and statistics (Cambridge University Press, Cmabridge).
Park, J. W. 2010. Global Economic Crisis and Stock Markets: Impact and Recovery. Parrino, R., and D. S. Kidwell. 2009. Fundamentals of corporate finance: Wiley. Patell, J. M. 1976. Corporate Forecasts of Earnings per share and stock price behavior:
Empirical tests Journal of accounting research 14 (2):246 - 276. Patell, J. M., and M. A. Wolfson. 1979. Anticipated information releases reflected in call
option prices. Journal of Accounting and Economics 1 (2):117-140. Peterson, P. P. 1989. Event studies: A review of issues and methodology. Quarterly
Journal of Business and Economics 28 (3):36-66. Pichardo, C., and F. Bacon. 2009. The Lehman Brothers Bankruptcy: A test of Market
Efficiency. Proceedings of Academy of Accounting and Financial Studies 14 (1):43 - 48.
Pindyck, R. S. 1984. Risk, inflation, and the stock market. American economic review 74 (3):334 – 351.
Poterba, J. M., and L. H. Summers. 1986. The Persistence of Volatility and Stock Market Fluctuations. American economic review 76 (5):1141-1151.
Raddatz, C. 2010. When the rivers run dry: liquidity and the use of wholesale funds in the transmission of the US subprime crisis: World Bank, 1-54.
Rendu de Lint, C. 2002. Risk profiles: how do they change when stock markets collapse? Journal of International Financial Markets, Institutions and Money 12 (1):59-80.
Ritter, J. R. 1991. The long-run performance of initial public offerings. Journal of finance 46 (1):3-27.
Samarakoon, L. P. 2011. Stock market interdependence, contagion, and the US financial crisis: The case of emerging and frontier markets. Journal of International Financial Markets, Institutions and Money 21 (5):724-742.
Schweitzer, R. 1989. How do stock returns react to special events? Business Review 8 (July/August):17-29.
Schwert, G. W. 1981. Using financial data to measure effects of regulation. Journal of law and Economics 24 (1):121-158.
Schwert, G. W. 1989. Why does stock market volatility change over time? Journal of Finance 44 (5):1115 – 1153.
Sharpe, W. F. 1964. Capital asset prices: A theory of market equilibrium under conditions of risk*. The Journal of Finance 19 (3):425-442.
Shin, J. 2005. Stock returns and volatility in emerging stock markets. International Journal of Business and Economics 4 (1):31-43.
Soydemir, G. 2000. International transmission mechanism of stock market movements: evidence from emerging equity markets. Journal of Forecasting 19 (3):149-176.
Syllignakis, M. N., and G. P. Kouretas. 2011. Dynamic correlation analysis of financial contagion: Evidence from the Central and Eastern European markets. International Review of Economics & Finance 20 (4):717-732.
117
Tai, C. S. 2007. Market integration and contagion: Evidence from Asian emerging stock and foreign exchange markets. Emerging Markets Review 8 (4):264-283.
Turner, C. M., R. Startz, and C. R. Nelson. 1989. A Markov model of heteroskedasticity, risk, and learning in the stock market. Journal of Financial Economics 25 (1):3-22.
Visaltanachoti, N., and R. Luo. 2009. Order imbalance, market returns and volatility: evidence from Thailand during the Asian crisis. Applied Financial Economics 19 (17):1391-1399.
Wang, J., G. Meric, Z. Liu, and I. Meric. 2009. Stock market crashes, firm characteristics, and stock returns. Journal of Banking & Finance 33 (9):1563-1574.
Wang, Z., A. M. Kutan, and J. Yang. 2005. Information flows within and across sectors in Chinese stock markets. The Quarterly Review of Economics and Finance 45 (4-5):767-780.
Whitelaw, R. F. 1994. Time variations and covariations in the expectation and volatility of stock market returns. Journal of finance 49 (2):515-541.
———. 2000. Stock market risk and return: An equilibrium approach. Review of Financial Studies 13 (3):521 - 547.
Wu, G. 2001. The determinants of asymmetric volatility. Review of Financial Studies 14 (3):837 - 859.
Xing, X., and J. S. Howe. 2003. The empirical relationship between risk and return: evidence from the UK stock market. International Review of Financial Analysis 12 (3):329-346.
118
APPENDIX A
1. Insert paper entitled “Sectoral and industrial performance during a stock market crisis” that has been published in “Economic Systems”.
2. Insert abstract from conference proceeding.
Sectoral and industrial performance during astock market crisis
Kumari Ranjeeni *
The University of the South Pacific, Laucala Campus, Suva, Fiji
1. Introduction
The non-strategic bankruptcy of Lehman Brothers (LBs)1 occurred in the United States (US)financial sector. The market considers a non-strategic bankruptcy to be bad news having an adverse
Economic Systems 38 (2014) 178–193
A R T I C L E I N F O
Article history:
Received 20 September 2013
Received in revised form 11 October 2013
Accepted 5 December 2013
Available online 30 March 2014
JEL classification:
G01
G11
G14
G21
G33
Keywords:
Lehman Brothers’ bankruptcy
Global financial crisis
Abnormal returns
Sectors
Trading strategy
A B S T R A C T
This paper investigates the impact of the news announcement of the
Lehman Brothers’ (LBs) bankruptcy on the performance of the New
York Stock Exchange (NYSE) sectors and financial industries. Based on
descriptive index level results, Bartram and Bodnar (2009) conclude
that the reaction of all sectors and industries was homogeneous
during the LBs’ bankruptcy and equity investors could not benefit
from diversification. Motivated by Narayan and Sharma’s (2011)
findings on firm and sector heterogeneity, this paper employs an
event study approach to further examine the sectoral and industrial
performance during the bankruptcy period. Daily data for a total of
481 firms is examined. The main contribution of this paper is that it
provides evidence that sectors behave heterogeneously during a
stock market crisis and the significant adverse impact from the LBs’
bankruptcy is discriminatory toward the financial sector and the
diversified financial industry, which were most exposed to LBs. This
paper proposes for investors to short sell those sector’s or industry’s
securities that are anticipated to be the most significantly adversely
affected from a particular negative news announcement.
� 2014 Elsevier B.V. All rights reserved.
* Tel.: +61 410199805.
E-mail address: [email protected] LBs was a distressed financial institution that filed for Chapter 11 bankruptcy with petition no. 308-13555 on 15 September,
2008 (see Dumontaux and Pop, 2012; Chong, 2011).
Contents lists available at ScienceDirect
Economic Systems
journal homepage: www.elsevier.com/locate/ecosys
http://dx.doi.org/10.1016/j.ecosys.2013.12.002
0939-3625/� 2014 Elsevier B.V. All rights reserved.
impact on the stock market performance (Coelho and Taffler, 2008). In addition, a crisis originatingfrom the financial sector is widely believed to be more contagious and risky than those in other sectorsof the economy (Kaufman, 1994). Evidence suggests that the LBs’ bankruptcy triggered incredibledeclines in index levels and escalated price volatility universally (Bartram and Bodnar, 2009; Chong,2011; Eichler et al., 2011; Samarakoon, 2011). The index levels at aggregate level proxy theperformance of the stock market firms and sectors. A disadvantage of examining index levelperformance is that all firms and sector constituents of the index are erroneously assumed to behomogeneous. Similarly, based on descriptive index level results, Bartram and Bodnar (2009) showthat the LBs’ bankruptcy had a homogenous impact on the performance of all sectors and industriesand that equity investors could not benefit from diversification. However, a recent study by Narayanand Sharma (2011) suggests that share market firms and sectors are heterogeneous in nature, with animplication that the LBs’ bankruptcy may not impose a significant adverse impact across all sectors.The question that therefore arises is: Did the LBs’ bankruptcy have a homogenous impact on the
performance of all sectors or was the adverse impact heterogeneous toward the financial sector and the
diversified financial industry which were most exposed to LBs? This research question has not beenaddressed so far. Therefore, this study takes the initiative of using a disaggregated approach toinvestigate whether the financial sector and the diversified financial industry were most significantlyadversely affected during the LBs’ bankruptcy. This paper also highlights the economic significance ofresults.
This study is motivated by four pioneering studies: Raddatz (2010), Bartram and Bodnar (2009),Dumontaux and Pop (2012) and Pichardo and Bacon (2009). Bartram and Bodnar (2009) provideevidence that the index returns of all sectors and industries were adversely affected during the LBs’bankruptcy (12 September 2008–27 October 2008). However, Bartram and Bodnar (2009) provideonly descriptive index level results. Therefore, no conclusions can be drawn on the significance (if any)of the impact of the LBs’ bankruptcy on the sectors and industries. On the other hand, Raddatz (2010),Dumontaux and Pop (2012) and Pichardo and Bacon (2009) used event study methodologies andsignificance tests to examine stock price returns during the LBs’ bankruptcy. Raddatz (2010)investigated the impact of the LBs’ bankruptcy on stock price returns of 662 individual banks across 44countries excluding the US. Dumontaux and Pop (2012) examined the contagion effects from the LBs’bankruptcy on the surviving large US financial institutions for the year 2008. Pichardo and Bacon(2009) examined abnormal returns of 15 investment firms, 9 of which had significant stakes in LBs, forthe period between 1 September 2008 and 27 October 2008.
Although Dumontaux and Pop (2012) and Pichardo and Bacon (2009) have used significance teststo examine abnormal return performance during the LBs’ bankruptcy, the investigation is conductedonly on US financial firms that are large2 and/or operating in the same industry as LBs. Likewise,Raddatz (2010) examined the performance of banks.
The absence of a thorough investigation of the impact of the LBs’ bankruptcy across sectors andfinancial industries is a gap in the literature. A thorough investigation is important for two reasons. (1)Stronger conclusions can be drawn on the pervasiveness of the impact from the LBs’ bankruptcy toascertain better insights on whether sectors and industries behave homogeneously or heteroge-neously during such volatile periods. (2) The disaggregated sectoral and industrial level analysis willenable the construction of profitable trading strategies. Consequently, this paper extends the extantliterature by employing an event study methodology and thoroughly investigating the impact of thenews announcement of the LBs’ bankruptcy on the performance of 10 sectors and all 4 financialindustries3 from the NYSE. All financial firms (irrespective of firm size) and financial industriesincluding real estate investment trusts (REITs) and insurances are examined. The event study analysisused in this paper is not only robust but also has strong theoretical foundations.4
2 ‘‘Large’’ is defined as firms ‘‘that reported total assets higher than US$ 1 billion in the last audited financial report before the
event date’’ (Dumontaux and Pop, 2012, p.8).3 An ‘‘industry’’ specifically refers to a group of firms involved in the same line of business operations, while a ‘‘sector’’
comprises a group of similar industries.4 Conversely, Bartram and Bodnar (2009) findings were only descriptive. The event study approach employed follows the
theory of market efficiency and supports Pichardo and Bacon (2009) on the efficiency of the US market during the LBs’
bankruptcy.
K. Ranjeeni / Economic Systems 38 (2014) 178–193 179
The main contributions of this paper are as follows. First, it provides evidence that sectors behaveheterogeneously during a crisis. During the LBs’ bankruptcy, the firms’ returns were affecteddifferently depending on their sectoral and financial industrial location. Second, this paper shows thatthe financial sector and the diversified financial industry, being the most exposed to LBs, were mostsignificantly adversely affected during the LBs’ bankruptcy. Third, the results suggest that investorscould short sell those sectors’ or industries’ securities that are anticipated to be the most significantlyadversely affected from a particular negative news announcement. This paper provides evidence thatthose investors who engaged in the short sale of financial sector and diversified financial securitiesover the [�4, 0] window would have benefited. The AAR from a short sale on day-4 followed by a shortcovering on day 0 is a significant (at the 1% level) 5.66% for the financial sector and 8.83% for thediversified financial industry.
The remainder of this paper is organized as follows. Section 2 develops the hypotheses. Next, themethodology and data used are discussed. This is followed by a presentation of the main findings ofthis study and robustness testing of the results in Sections 4 and 5, respectively. In Section 6, theeconomic significance of the results is discussed. The final section provides concluding remarks.
2. The relationship between Lehman Brothers, the financial sector and the diversified financialindustry
Gorban et al. (2010) suggest that, in times of crisis, correlation and volatility (risk) increasesimultaneously amongst firms and sectors.5 Acharya et al. (2009) show the LBs’ bankruptcy havingsignificant systematic risk and Bartram and Bodnar (2009) find increased correlation across the USsectors following the LBs’ bankruptcy. However, there is also evidence to suggest that firms with directcredit or investment exposure to LBs suffered more than those not directly exposed to LBs(Dumontaux and Pop, 2012; Pichardo and Bacon, 2009; Chakrabarty and Zhang, 2010).
Bartram and Bodnar (2009) find all US sector indices suffering a decline in returns and havingincreased volatility during the LBs’ bankruptcy. However, the US financial sector had a higher standarddeviation than the non-financials (Bartram and Bodnar, 2009). LBs, being a constituent of the USfinancial sector, is likely to associate more with firms from the financial sector as opposed to firmsfrom other sectors. Additionally, LBs had heavily invested in securities linked to the US subprimemortgage market. The financial industries, namely banks, diversified financials, insurances and REITs,were involved in subprime mortgages. It follows that, since these financial industries are constituentsof the financial sector, LBs was most associated with the financial sector in comparison to othersectors. In Table 1, the correlation between the LBs’ price returns and each of the sector’s index returnsis provided for the periods before the global financial crisis (GFC) (1994–2006) and during the GFC(2007–15 September 2008).
Table 1 shows that the financial sector was significantly positively correlated with LBs at the 1%level. LBs was most strongly correlated with the financial sector (the correlation coefficient was 0.63)during the GFC. This drives the motivation for the need to examine whether the LBs’ bankruptcy mostsignificantly adversely influenced the financial sector performance.
Hypothesis 1. Financial sector performance is the most significantly adversely affected during the LBs’bankruptcy.
Bartram and Bodnar (2009) find all US financial sector industries to be adversely affected duringthe LBs’ bankruptcy. Studies reveal that during the GFC, the subprime collateralized debt obligation(CDO) market was a major driver in the transmission of volatility to the large complex financialinstitutions’ (LCFIs) equity returns (Longstaff, 2010; Calice, 2011). Calice (2011) finds the largecomplex financial institutions’ equity returns to be positively correlated with the subprime asset-backed CDO market index returns. Raddatz (2010) finds that globally and within countries, thosebanks relying heavily on non-deposit sources of funds experienced a substantial fall in stock returns.Dumontaux and Pop (2012) examine the performance of disaggregated large banks and the
5 How? Since share market sectors are found to be integrated, a shock from any sector may influence other sector returns
(Wang et al., 2005).
K. Ranjeeni / Economic Systems 38 (2014) 178–193180
‘‘non-bank’’ diversified financial industry and find contagion effects discriminatory toward the biggestfinancial firms in the financial sector. The authors indicate that the impact of the LBs’ bankruptcy waslimited to the largest financial firms and significantly affected surviving ‘‘non-bank’’ financial servicesoffering the same products (mortgage and specialty finance, investment services and diversifiedfinancial services firms) with similar financial conditions, risk profiles and characteristics, and that thecomponents of the disaggregated diversified financial industry’s performance (investment servicesand diversified financial services firms) were most significantly adversely affected from the LBs’bankruptcy. In support, Pichardo and Bacon (2009) find the performance of investment firms (whichare constituents of the diversified financial industry), including those that had significant stakes in LBs,to be greatly adversely affected from the LBs’ bankruptcy on and around the event date (15thSeptember 2008). This suggests that LBs, being an investment bank and a constituent of the diversifiedfinancial industry, would have been more closely associated with other diversified financial firms thanfirms from the other three financial industries (banks, insurances and REITs). This motivates theinvestigation of whether the diversified financial industry amongst other financial industries was themost significantly adversely affected from the LBs’ bankruptcy.
Hypothesis 2. The diversified financial industry’s performance is the most significantly adversely affectedduring the LBs’ bankruptcy.
3. Methodology and data
3.1. Methodology
This paper employs an event study approach to examine the impact of the news announcement ofthe LBs’ bankruptcy on the share price returns performance of NYSE sectors and financial industries.Other studies that have used event study methodology to examine single day events includeKryzanowski et al. (1995), Dumontaux and Pop (2012), Pichardo and Bacon (2009), Raddatz (2010)and Mio and Fasan (2012). Kryzanowski et al. (1995) used it to examine the impact of the Canadianstock market crash of 1987 on the performance of screen-sorted portfolios’ abnormal returns,volatility, and residual risk premium, while Dumontaux and Pop (2012), Pichardo and Bacon (2009)and Raddatz (2010) used it to examine the impact of the LBs’ bankruptcy on the performance offinancial firms. Mio and Fasan (2012) used event study methodology to examine whether corporatesocial performance had any impact on corporate financial performance due to the LBs’ bankruptcy. Theauthors analyzed the S&P 500 market index’s constituent non-financial companies’ stock prices priorto and during the LBs’ bankruptcy announcement.
Table 1Correlation results between LBs and each of the 10 sectors.
Sectors Before GFC (1994–2006) During GFC (2007–15 September 2008)
FIN 0.09*** 0.63***
IND 0.14*** 0.08
CS 0.03** 0.1**
MAT 0.04* 0.14***
UTI 0.03 0.15***
TS 0.02 0.23***
IT 0.07*** 0.15***
HC 0.05** 0.14***
ENG 0.02 0.24***
CD 0.07*** 0.21***
Notes: This table provides the correlation between the LBs’ price returns and each of the sectors index returns
for the periods before the GFC (1994–2006) and during the GFC (2007–15 September 2008).* Significant at 10% level.** Significant at 5% level.*** Significant at 1% level.
K. Ranjeeni / Economic Systems 38 (2014) 178–193 181
Furthermore, this paper uses an event study approach to examine both hypotheses as the USmarket is semi-strong form efficient6 (see MacKinlay, 1997; Pichardo and Bacon, 2009). Therefore,NYSE security returns are anticipated to be instantaneously reflective of the effects of the newsannouncement of the LBs’ bankruptcy on the firm’s value. This paper uses daily returns to conduct theevent study because Brown and Warner (1985) find the use of daily returns to be more powerful thanmonthly returns for event study purposes. At the initial stage of conducting an event study, the eventalong with the window over which the security prices will be observed needs to be defined (Campbellet al., 1997). In this paper, the event date is defined as Monday 15 September 2008, the newsannouncement of the LBs’ bankruptcy (Dumontaux and Pop, 2012). This is represented by day ‘‘0’’ inthis paper. Prior research using 15 September 2008 as the event date to examine the impact from theLBs’ bankruptcy (on the performance of financial firms) include Pichardo and Bacon (2009),Dumontaux and Pop (2012) and Raddatz (2010). In this paper, the choice of 15 September 2008 as theevent date is motivated by the following reasons. (1) Although LBs reported an enormous loss of $4billion on 10 September 20087 (Bartram and Bodnar, 2009), this is not chosen as the event date as theevent of 15 September 2008 stands out due to the extent and pace of its alarming consequences(Raddatz, 2010). On 15 September 2008, LBs had total assets of $639 billion, making it the largestfailure in US history. This stimulates the need to examine the stock market reaction to the catastrophicnews of the immense failure of the largest fixed interest security dealer, LBs. (2) It is the newsannouncement of the LBs’ bankruptcy on 15 September 2008 that triggered the GFC, making it anevent worth investigating. Despite the ongoing mortgage and banking crisis since early 2007, thebankruptcy of LBs on 15 September 2008 marks the ‘‘real’’ anticlimax era in the stock marketperformance (Bartram and Bodnar, 2009). ‘‘September 15, 2008 has been proclaimed Wall Street’sworst day in seven years. The Dow Jones Industrial average lost more than 500 points, more than 4%,which is the steepest fall since the day after the September 11th attacks’’ (Pichardo and Bacon, 2009).(3) The findings of this paper will be an extension to the findings of Pichardo and Bacon (2009),Dumontaux and Pop (2012) and Raddatz (2010) with the use of a comprehensive sample selection.
Additionally, in order to control for multiple negative news announcements around the LBs’bankruptcy impairing the variance estimates of abnormal return,8 a shorter event window of 5 tradingdays is observed as opposed to the typical lengths of 21–121 days (Peterson, 1989). Two such negativenews announcements were the placement of Fannie Mae and Freddie Mac into governmentconservatorship by the Federal Housing Finance Agency and the bailout of the American InternationalGroup (AIG). Fannie Mae and Freddie Mac were put into government conservatorship by the FederalHousing Finance Agency on Sunday, 7 September 2008 (Bartram and Bodnar, 2009) and thesubsequent day was 5 trading days prior to the LBs’ bankruptcy. On the other hand, AIG was bailed outon the day subsequent to the event date (Bartram and Bodnar, 2009). In order to control for theaforementioned negative news announcements, 5 trading days prior to LBs’ bankruptcy and the dayssubsequent to the event date are not included in the event window.
The negative news announcements on LBs started 4 trading days prior to the bankruptcy (Bartramand Bodnar, 2009). On -4 trading days (9 September 2008) ‘‘Lehman Brothers shares plummet tolowest level on Wall Street in more than a decade’’ (Bartram and Bodnar, 2009, p. 1280).9 On -3 tradingdays (10 September 2008), ‘‘Lehman Brothers puts itself up for sale after reporting a $4 billion loss andsays it will spin off its troubled commercial real estate assets’’ (Bartram and Bodnar, 2009, p. 1280). On-1 trading days (12 September 2008), ‘‘with Lehman Brothers facing collapse, US officials struggle to
6 Pichardo and Bacon (2009) find the US market semi-strong form efficient during the LBs’ bankruptcy.7 This papers’ event window of [�4, 0] trading days incorporates the effects of LBs’ huge loss reported on 10 September 2008.8 As a robustness test, a longer 61-day event period as used by Pichardo and Bacon (2009) is also analyzed. However, during
the 61-day event period, there were multiple negative news announcements unrelated to LBs. Negative news announcements
on LBs start from 4 trading days prior to the LBs’ bankruptcy (see Bartram and Bodnar, 2009) and the CAAR results (which have
not been reported to conserve space) show that the financial sector, being the most exposed to LBs, had adverse cumulative
returns from days �4 to 0. This supports the choice of this study’s 5 days event window.9 The sharp decline in LBs’ shares indicates that after Fannie Mae and Freddie Mac were put into government conservatorship
by the Federal Housing Finance Agency on Sunday, 7 September 2008 (Bartram and Bodnar, 2009), the market participants
anticipated that LBs would be the next troubled investment bank, which greatly affected the investor confidence amongst LBs
shareholders.
K. Ranjeeni / Economic Systems 38 (2014) 178–193182
find a buyer for the distressed investment bank’’ (Bartram and Bodnar, 2009, p. 1280). On the eventday, ‘‘12.30am EST: Lehman Brothers Holdings Incorporated files for Chapter 11 bankruptcyprotection. SEC Filing’’ (Bartram and Bodnar, 2009, p. 1281). Since negative news announcements onLBs dominated the timeline of events for the period �4 trading days to the event date, (Bartram andBodnar, 2009), the impact of the LBs’ bankruptcy is analyzed for an event window of [�4, 0] tradingdays.
Event study approaches typically use an estimation period immediately prior to the event windowin predicting normal period returns. Since the estimation period immediately prior to the eventwindow of [�4, 0] trading days was exposed to a large number and a high frequency of events causedby the GFC, it is inappropriate to use an estimation period during this time to predict normal periodreturns (see Dumontaux and Pop, 2012). Following Raddatz (2010), an estimation period of 150 daysprior to 30 June 2007 (22/11/2006–29/6/2007) is used to predict normal period returns that wouldhave eventuated in the absence of the GFC. The period before 30 June 2007 is assumed to be normalperiod. Although the GFC started in August 2007 (Dumontaux and Pop, 2012), in July 2007 two BearStearns hedge funds specializing in subprime debt filed for Chapter 15 bankruptcy and CountrywideFinancial, one of the biggest mortgage originators in the US, warned of ‘‘difficult conditions’’ (Raddatz,2010).
According to MacKinlay (1997), the event study methodology prevalent today is essentially thesame as that introduced in the late 1960s in the seminal studies of Ball and Brown (1968) and Famaet al. (1969). Additionally, MacKinlay (1997) warrants consideration for the use of multi-factor modelsin conducting event studies for firms that are all members of the same industry. Therefore, this paperemploys the Fama and French (1993) three factor model10 to examine the performance of securityreturns at sector and financial industry level. In the estimation period, each of the firm’s daily excessreturns at time t is regressed on the market factor, size factor and book-to-market factor:
Rit � R ft ¼ ai þ biðRmt � R ftÞ þ siSMBt þ hiHMLt þ eit ; (1)
where Rit is the actual realized return for firm i at time t; Rft is the daily return on one-month Treasurybills; Rmt is the value-weight return on all NYSE, AMEX, and NASDAQ stocks (from CRSP); SMBt is thedifference between the average return on the three small portfolios and the average return on thethree big portfolios; HMLt is the difference between the average return on the two value portfolios andthe average return on the two growth portfolios.
The estimated coefficients, the risk free rate and the realized market, size and book-to-market riskpremiums during the event window are used to compute the expected return for the event window[�4, 0] trading days. Following this, the actual realized return (Rit) during the event window iscompared with the predicted normal return based on the Fama and French (1993) three factor modeland the difference is known as abnormal return (ARit):
ARit ¼ Rit � EðRitÞ; (2)
where ARit is the abnormal return for security i at time t, Rit is the actual realized return for security i attime t, and E(Rit) is the expected normal period return for security i at time t.
A positive (negative) abnormal return implies that the LBs’ bankruptcy had a positive (adverse)impact on the performance of the security’s return. ‘‘The event study literature typically privileges theanalysis of cumulative returns because the cumulative impact of the events is easier to visualize’’(Raddatz, 2010, p. 11). Also, negative news announcements on LBs dominated the event window [�4,0] trading days (Bartram and Bodnar, 2009). Therefore, in line with conventional event studyreporting, this paper focuses on the evolution of average cumulative returns over the event window of[�4, 0] trading days. However, the daily average abnormal returns (AAR) on the event date are alsoreported.
10 The Fama et al. (1969) market model and market-adjusted returns model are employed to check the robustness of the
results.
K. Ranjeeni / Economic Systems 38 (2014) 178–193 183
The AAR for each of the sectors and financial industries on any particular day t is computed asfollows:
AARGt ¼PN
i¼1 ARit
N(3)
where N is the number of firms in the sector/financial industry, ARit is the abnormal return for securityi at time t, and AARGt is the average abnormal return for the sector/financial industry.
The AARGt is then accumulated through time to generate the cumulative average abnormal return(CAAR). The CAAR for each sector and financial industry for any particular day t is computed as follows:
CAARGt ¼XT
t¼T1
AARGt (4)
where CAARGt is the cumulative abnormal return for sectors/financial industries over a series of timeduring the event window of [�4, 0] trading days, T1 is the first period in which the AARGt areaccumulated, and T2 is the last period in which the AARGt are accumulated.
Finally, the ordinary cross-sectional approach (see Brown and Warner, 1980) is employed tocompute the test statistics for the AAR and CAAR.
3.2. Data
This study uses a disaggregated approach to examine the performance of NYSE sectors andfinancial industries around the news announcement date of the LBs’ bankruptcy (referred to as theevent date). The 9 March 2011 S&P 500 composition firm data list, which contained information oncompany name and ticker symbol, was downloaded from Standard and Poor’s website.11 Each of the
Table 2Sample dataset.
Panel A: Sample selection procedures for each of the 10 sectors
Sectors Total firms Number of firms removed Ticker symbol of firms removed Firms remaining
CD 79 2 SNI, TWC 77
CS 41 4 DPS, LO, MJM, PM 37
Eng 39 1 SE 38
Fin 82 3 BRKb, CMF, DFS 79
HC 51 2 CFN, COV 49
Inds 62 - – 62
IT 72 3 MMI, V, TDC 69
Mats 30 - – 30
Uti 35 2 PEG, QEP 33
TS 8 1 PCS 7
Total 499 18 481
Panel B: Sample size for the financial industries
Industries Sample size
Banks 20
Diversified financials 21
Insurance 20
REIT 16
Total 771
Notes: This table details the sample used in this paper. Any firm with missing data during the estimated period or the event
period was removed from the sample to avoid biasness in the results. The final sample consists of a total of 481 firms grouped
into 10 sectors and four financial industries.1 Amongst the 79 financial firms, only 77 could be classified into the four financial industries.
11 LBs was listed on the NYSE and traded as part of the S&P 500 index prior to its bankruptcy. This motivated the choice of using
the S&P 500 index data to analyze the impact of the LBs’ bankruptcy at the sector and financial industry levels.
K. Ranjeeni / Economic Systems 38 (2014) 178–193184
companies’ ticker symbols was used to download their respective closing share price data from theCentre for Research in Security Prices database. Following this, each of the firms was distributed into10 sectors and 4 financial industries according to their corresponding Global Industry ClassificationStandards sector classification (GICS). The 10 sectors are Consumer Discretionary (CD), ConsumerStaples (CS), Energy (Eng), Financials (Fin), Health Care (HC), Industrials (Ind), Information Technology(IT), Materials (Mat), Utilities (Uti), and Telecommunication Services (TS).12 Table 2 panel A details thesample selection procedures. Any firm with missing data during the estimated period or the eventperiod was removed from the sample to avoid bias in the results. As a result, the sample used in thispaper consists of a total of 481 firms13: 77 firms in the CD sector, 37 in CS, 38 in Eng, 79 in Fins, 49 inHC, 62 in Inds, 69 in IT, 30 in Mats, 33 in Uti, and 7 in the TS sector. The four financial industries are:Banks, Diversified Financials, Insurances and Real Estate Investment Trusts (REIT). There were 20 firmsin Banks, 21 firms in Diversified Financials, 20 firms in Insurance and 16 firms in REIT. The daily Famaand French factors data source is Wharton Research Data Services.
4. Discussion of findings
4.1. Findings for the 10 sectors
In this section, event study results for the 10 sectors are presented. The abnormal returnperformances of all 10 sectors were analyzed to determine whether the financial sector identified asthe epicenter for the GFC (see Bartram and Bodnar, 2009) was the most significantly adversely affectedduring the LBs’ bankruptcy. The findings of t-tests on the analysis of AAR and CAARs per sector appearin Tables 3 and 4 respectively.
The main findings on the performance of the sectors based on the AAR and CAARs results aresummarized as follows. On the event date, eight sectors (CD, CS, Eng, Fin, HC, IT, TS, Uti) weresignificantly affected, while two sectors (Ind and Mat) were insignificantly affected. Sectors aresignificantly affected at the 1% level with the exception of Eng, Fin and HC, which have significance atthe 5% level. Amongst the significantly affected sectors, the LBs’ bankruptcy had a positive impact onfive sectors (CD, CS, HC, TS, Uti) and an adverse impact on only three sectors (Eng, Fin, IT). The Finsector AAR (�2.11%) is most adversely affected, followed by IT (�1.58%) and Energy (�1.45%). TheCAAR for the [�4, 0] trading days event window show that eight sectors were significantly affected(CS, Eng, Fin, HC, Ind, IT, Mat, Uti) while two were insignificantly affected (CD, TS) by the LBs’bankruptcy. Sectors are significantly affected at the 1% level with the exception of HC and Ind, whichhave significance at the 5% and 10% levels respectively. Amongst the eight significantly affectedsectors, the LBs’ bankruptcy had an adverse impact on only two sectors (Fin and IT). Fin is the mostadversely affected (�4.8% at 1% level of significance) followed by IT (�2.95% at 1% level). The AAR andCAAR results provide evidence in support of sector heterogeneity (see Narayan and Sharma, 2011) byshowing that not all sectors’ performance was significantly adversely affected by the LBs’ bankruptcy.During the LBs’ bankruptcy, the returns of sectors were affected differently.
Overall, the results provide evidence in support of my proposition that the financial sector is themost significantly adversely affected by the LBs’ bankruptcy during the event window of [�4, 0]trading days. On the event date, the Fin sector has the highest negative AAR of �2.11%, significant atthe 5% level. Additionally, the CAAR results show that the financial sector is the only sectorconsistently adversely affected at the 1% level of significance over the multi day intervals of [�4, �3],[�4, �2], [�4, �1] and [�4, 0] trading days. Consequently, the Fin sector has the highest CAAR of�4.8%, significant at the 1% level for the window of [�4, 0] trading days. This implies that due to thestrong degree of correlation between LBs and the Fin sector14 (see Table 1), negative LBs newsannouncements had the most significant adverse impact on Fin sector performance. This also provides
12 Since the Consumer Services sector had only one firm, it was dropped from the analysis and only 10 sectors are analyzed in
this study.13 Under the MM, the sample consisted of 488 firms: 78 firms in the CD sector, 37 in CS, 39 in Eng, 81 in Fins, 50 in HC, 62 in
Inds, 70 in IT, 30 in Mats, 33 in Uti, and 8 in the TS sector.14 LBs was significantly correlated with the Fin sector before the GFC (1994–2006) and during the GFC (2007–15 September
2008). LBs was most significantly correlated with the Fin sector during the GFC.
K. Ranjeeni / Economic Systems 38 (2014) 178–193 185
Table 3Findings of t-tests on the analysis of AAR per sector based on the Fama and French (1993) three factor model.
Days Consumer
discretionary
Consumer
staples
Energy Financials Health
care
Industrials Information
technology
Materials Telecommunication
services
Utilities
0 0.0117***
(3.5356)
0.016***
(5.5419)
�0.0145**
(�2.1923)
�0.0211**
(�2.4775)
0.0087**
(2.4636)
0.0013
(0.3273)
�0.0158***
(�4.3348)
0.0051
(0.9296)
0.0597***
(10.8131)
0.0303***
(4.6216)
Notes: This table reports AAR for 10 sectors on the event date (15 September 2008) based on the Fama and French (1993) three factor model. t-Statistics are reported in parentheses. A
positive sign implies a positive impact from the event, while a negative sign implies an adverse impact from the event.
t-Statistics are based on the ordinary cross-sectional approach for significance testing.** Significant at 5% level.*** Significant at 1% level.
K.
Ranjeen
i
/
Economic
System
s
38
(2014)
178–193
186
evidence in support of a semi-strong form of market efficiency by showing a direct relationshipbetween the negative news announcement on LBs and the price return performances of the Fin sector.
4.2. Findings for the financial industries
In this section, event study results for the financial industries are presented. Findings of t-tests onthe analysis of AAR and CAARs per financial industry appear in Tables 5 and 6 respectively.
The main findings are summarized as follows. Further to the study by Narayan and Sharma (2011),this study provides evidence on industry heterogeneity during the LBs’ bankruptcy. Although theperformance of all financial industries was adversely affected on the event date (depicted by negativeAAR) and over the window of [�4, 0] trading days, not all financial industries’ performance wassignificantly adversely affected. On the event date, only diversified financials and REIT performanceswere significantly adversely affected. The diversified financials AAR (�3.53%) is significantly affectedat the 1% level of significance, followed by REIT (�2.54%) at the 5% level of significance. Additionally,the CAAR results for the [�4, 0] window show that only diversified financials was significantlyadversely affected at the 1% level. The other three financial industries were insignificantly affectedduring the [�4, 0] window. This implies that although all financial industries were adversely affectedduring the LBs’ bankruptcy, the adverse impact was not significant for all financial industries. Insupport, the findings of Dumontaux and Pop (2012) ‘‘suggest that the observed contagious effectswere rational/discriminating rather than panic-driven/undifferentiated and tend to weaken the casefor the bailout of Lehman Brothers’’.
Overall, the AAR and CAARs results show that the significant adverse impact was discriminatorytoward the diversified financial industry, which was most exposed to LBs. This provides evidence insupport of my proposition that the diversified financial industry’s performance was the mostsignificantly adversely affected during the LBs’ bankruptcy. The diversified financial industry’sperformance is the most significantly adversely affected at the 1% level of significance not only on theevent date (depicted by AAR) but also on a cumulative basis over the multi day intervals of [�4, �3],[�4, �2], [�4, �1] and [�4, 0] trading days. This also provides evidence in support of a semi-strongform of market efficiency by showing a direct relationship between the negative news announcementon LBs and the price return performances of the diversified financial industry. Diversified financials isthe only industry having consistent negative CAAR significant at the 1% level over the window of [�4,0] trading days. This implies that since LBs was an investment bank, negative news announcements onLBs greatly affected the confidence among the diversified financial industry’s investors in comparisonto other financial industries. In support, based on the market-adjusted model, Dumontaux and Pop(2012) find the components of disaggregated diversified financial performance (investment servicesand diversified financial services firm) to be most significantly adversely affected by the LBs’bankruptcy. Therefore, this study’s findings using the Fama and French (1993) three factor model anda larger sample representation15 complement those of the study by Dumontaux and Pop (2012).
5. Robustness test
5.1. The market model
The Fama et al. (1969) market model is the most widely used approach for measuring abnormalreturns (for related studies, see Brown and Warner, 1985; Peterson, 1989; MacKinlay, 1997; Lee andConnolly, 2010; Chen and Lai, 2013). The market model (MM) is a more robust approach to event studyin comparison to other statistical (Brown and Warner, 1985; Ritter, 1991; MacKinlay, 1997), non-regressive (Cable and Holland, 1996, 1999) and economic approaches (MacKinlay, 1997). According toMacKinlay (1997), the OLS model is generally an efficient and consistent means of estimating the MMparameters. The S&P 500 index is used as the market index (Rmt). In the estimation period (150 daysprior to 30 June 2007), each of the firm’s share price returns at time t is regressed on the concurrent
15 This study examines all four financial industries, namely, banks, diversified financials, REITs and insurances, from NYSE. On
the other hand, Dumontaux and Pop (2012) examined only disaggregated bank and non-bank financial institutions.
K. Ranjeeni / Economic Systems 38 (2014) 178–193 187
Table 4Findings of t-tests on the analysis of CAARs per sector based on the Fama and French (1993) three factor model.
Window Consumer
discretionary
Consumer
staples
Energy Financials Health
care
Industrials Information
technology
Materials Telecommunication
services
Utilities
�4 0.0028
(0.9834)
0.0178***
(5.9099)
�0.035***
(�6.6946)
�0.0156***
(�4.6643)
0.0082**
(2.5718)
�0.0049
(�1.2847)
�0.0076**
(�2.5158)
�0.0102
(�1.5339)
�0.0017
(�0.4111)
0.0159***
(3.1952)
[�4, �3]
�0.0027
(�0.7577)
0.0125***
(4.0731)
0.0042
(0.6454)
�0.0267***
(�5.819)
0.0052
(1.3986)
�0.0009
(�0.2413)
�0.0091**
(�2.3937)
0.0046
(0.6647)
�0.0117
(�1.5546)
0.0219***
(5.3071)
[�4, �2]
�0.0001
(�0.0247)
0.0138***
(3.8185)
0.005
(0.7184)
�0.0265***
(�5.0299)
0.0107**
(2.0333)
0.0071
(1.5898)
�0.0121**
(�2.5139)
0.0073
(1.0675)
�0.0273**
(�3.6967)
0.0148***
(2.7972)
[�4, �1]
�0.0104**
(�2.5062)
0.0134***
(3.2816)
0.0412***
(5.3947)
�0.027***
(�3.1821)
0.0078
(1.458)
0.014***
(2.6663)
�0.0137***
(�2.7257)
0.0355***
(4.9712)
�0.0352**
(�3.3795)
0.0265***
(5.6285)
[�4, 0] 0.0013
(0.2183)
0.0294***
(5.5845)
0.0266***
(2.7652)
�0.048***
(�2.9934)
0.0165**
(2.384)
0.0152*
(1.9588)
�0.0295***
(�4.924)
0.0406***
(4.4325)
0.0244
(1.711)
0.0568***
(5.6677)
Notes: This table reports CAARs for 10 sectors based on the Fama and French (1993) three factormodel. t-Statistics are reported in parentheses. A positive sign implies a positive impact from
the LBs’ bankruptcy, while a negative sign implies an adverse impact.
t-Statistics are based on the ordinary cross-sectional approach for significance testing.* Significant at 10% level.** Significant at 5% level.*** Significant at 1% level.
K.
Ranjeen
i
/
Economic
System
s
38
(2014)
178–193
188
S&P 500 index returns. The estimated alpha ( ai) and beta (bi) coefficients from the regression are usedto compute the expected normal period return (E(Rit)) for each security at time t during the eventwindow [�4, 0]. The abnormal return for each security at time t is computed using Eq. (2).
The standardized cross-sectional test proposed by Boehmer et al. (1991)16 is used to examine thestatistical significance of the AR’s results. Why? In this paper, the sample of securities is analyzed atsector and financial industry levels. Therefore, there could be a high degree of cross-sectionaldependence in the ARs arising from event date clustering.17 Brown and Warner (1985) state that ifsecurities are randomly selected in a sample, the one factor MM (as used in this study) providessufficient adjustment for dependence. However, ‘‘if instead the securities came from the sameindustry group (as in this study), with clustering there could be a higher degree of cross-sectionaldependence in market model excess returns (abnormal returns) and measurable misspecification’’(Brown and Warner, 1985, p. 22). Nevertheless, Boehmer et al. (1991) provide evidence that thestandardized cross-sectional test approach is robust18 in the presence of event date clustering as well
Table 5Findings of t-tests on the analysis of AAR per financial industry based on the Fama and French (1993) three factor model.
Days Banks Diversified financials Insurance REIT
0 �0.0071 (�0.7228) �0.0353*** (�3.4823) �0.0176 (�0.5942) �0.0254** (�2.6058)
Notes: This table reports AAR for financial industries on the event date (15 September 2008) based on the Fama and French
(1993) three factor model. t-Statistics are reported in the parentheses. A positive sign implies a positive impact from the LBs’
bankruptcy while a negative sign implies an adverse impact.
t-Statistics are based on the ordinary cross-sectional approach for significance testing.** Significant at 5% level.*** Significant at 1% level.
Table 6Findings of t-tests on the analysis of CAARs per financial industry based on the Fama and French (1993) three factor model.
Window Banks Diversified financials Insurance REIT
�4 �0.0131*** (�3.0272) �0.0325*** (�4.7899) �0.0133 (�1.4931) �0.0021 (�0.7601)
[�4, �3] �0.037*** (�4.5893) �0.0438*** (�6.1192) �0.0131 (�1.0855) �0.0053 (�0.8615)
[�4, �2] �0.0265** (�2.5814) �0.0467*** (�5.0415) �0.0236* (�1.8252) �0.0001 (�0.0205)
[�4, �1] �0.0049 (�0.5209) �0.0578*** (�5.34) �0.0449 (�1.6435) 0.0127 (1.7039)
[�4, 0] �0.012 (�0.7719) �0.0932*** (�4.9774) �0.0625 (�1.1069) �0.0127 (�0.9101)
Notes: This table reports CAARs for financial industries based on the Fama and French (1993) three factor model. t-Statistics are
reported in the parentheses. A positive sign implies a positive impact from the LBs’ bankruptcy, while a negative sign implies an
adverse impact.
t-Statistics are based on the ordinary cross-sectional approach for significance testing.* Significant at 10% level.** Significant at 5% level.*** Significant at 1% level..
16 The Boehmer et al. (1991) standardized cross-sectional test is a hybrid formed by combining the two approaches of
standardized residuals (Patell, 1976) and the ordinary cross-sectional method (see Brown and Warner, 1980).17 This refers to securities of a sample suffering from other close or simultaneous spacing of events during the event period
apart from the event under investigation (Brown and Warner, 1980). For example, Foster (1980) and Schwert (1981) discuss
government regulation or mandated accounting procedures simultaneously affecting the performance of numerous securities
under examination during a particular event.18 The standardized cross-sectional test approach adjusts for forecasted errors from the estimation period. Such an adjustment
controls for the misspecification that could result from even minor increases in variances caused by an event causing too many
false rejections of the null hypothesis of no abnormal return when it is in fact true (see Boehmer et al., 1991). Boehmer et al.
(1991) provide evidence that during the existence of event induced variances, the standardized cross sectional test performs
better than other commonly used approaches (ordinary cross-sectional test, standardized residuals test, sign tests and method
of moments approach) of significance testing. Accordingly, this study employs the standardized cross-sectional test to ensure
that the rejection of the null hypothesis is done accurately.
K. Ranjeeni / Economic Systems 38 (2014) 178–193 189
as event induced variances.19 In support, Harrington and Shrider, 2007 show that the standardizedcross-sectional test statistic suggested by Boehmer et al. (1991) is a good candidate for testing thesignificance of ARs.
The results20 are consistent with the main findings of this paper and show that the financial sectorand the diversified financial industry are most significantly adversely affected during the eventwindow [�4, 0].
5.2. Estimation period before the event period
Based on the Fama et al. (1969) market model, this paper also estimated normal period returns forall 10 sectors for a longer event window of [�30, +30] trading days (1/8/2008–23/1/2009) and anestimation window of 150 days (27/12/2007–31/7/2008) immediately before the event period. Theresults show that on the event date, the financial sector AAR is positive and insignificant (0.30%). Thisimplies that the estimation period returns immediately prior to the event period are highlycontaminated by the volatility from negative news announcements during the GFC. Therefore, theyare not a good estimation parameter to predict normal period returns during the LBs’ bankruptcy. Thissupports this paper’s and Raddatz’s (2010) approach of using an estimation period of 150 days prior to30 June 2007 to forecast normal period returns that would have eventuated in the absence of the GFC.
5.3. Market-adjusted returns model
As a further diagnostic check, the market-adjusted returns model is employed to estimateabnormal returns for each of the 10 sectors and 4 financial industries during the event window [�4, 0]trading days. Brown and Warner (1985) find that for daily data, the market-adjusted return modelworks reasonably well in examining the impact of an event and is also powerful in the presence ofevent date clustering. The market-adjusted return model is used in scenarios where ‘‘it is not feasibleto have a pre-event estimation period for the normal model parameters’’ (MacKinlay, 1997, p. 18). Asdiscussed in the methodology section and Section (5.2, since the GFC was a period characterized byhigh levels of volatility, an estimation period immediately prior to the LBs event provided impreciseresults. Accordingly, the market-adjusted return model is used to check the robustness of this paper’sfindings because it does not depend on past returns to forecast normal period returns (Dumontaux andPop, 2012).
Under the market-adjusted return model, the normal period return is the current event periodmarket return. The statistical significance of the AAR and CAARs are tested by employing the ordinarycross-sectional test approach (see Brown and Warner, 1985).
Overall, the results21 support the main findings of this study and show that the financial sector andthe diversified financial industry are the most significantly adversely affected during the eventwindow [�4, 0].
6. Economic significance
This paper has established three things. (1) The impact of the LBs’ bankruptcy varied across sectorsand financial industries. (2) The financial sector and the diversified financial industry, which were themost exposed to LBs, were most significantly adversely affected by the LBs’ bankruptcy. (3) The USmarket is semi-strong form efficient as there was a direct relationship between negative newsannouncements on LBs and the performance of the financial sector and diversified financial industryreturns over the event window [�4, 0]. The question that then follows is how an investor can utilizethese findings to devise a profitable trading strategy. This paper proposes for investors to short sell
19 There is evidence that the variance of stock returns increases for the days immediately around events, such as earnings
announcements (e.g., Beaver, 1968; Patell and Wolfson, 1979). Brown and Warner (1985) illustrate how variance increase can
cause misspecification of hypothesis tests conducted using standard event study procedures.20 To conserve space, only the main findings are discussed. However, the results can be made available upon request.21 To conserve space, only the main findings are discussed. However, the results can be made available upon request.
K. Ranjeeni / Economic Systems 38 (2014) 178–193190
those sectors or industry’s securities that are anticipated to be the most significantly adverselyaffected by a particular negative news announcement. In this paper, an average buy-and-holdabnormal return (BHAR) technique is used to test this proposition on the financial sector and thediversified financial industry’s securities.
Following Trueman et al. (2003), the average BHAR for the financial sector and diversified financialindustry’s security is computed by firstly compounding each security’s raw returns over the [�4, 0]event window. Subsequently, the S&P 500 index (market index) is compounded over the [�4, 0]window and subtracted from each of the security’s compounded raw returns. Lastly, the ARs areaveraged across all securities within the financial sector and the diversified financial industry. The teststatistics for the average BHAR over the [�4, 0] window is based on its corresponding cross-sectionalstandard error (see Barber and Lyon, 1997; Trueman et al., 2003). The negative of the average BHARover the [�4, 0] window provides the AAR from short selling during this window.
The findings on the AAR from short selling for the financial sector and the diversified financialindustry’s security are presented in Table 7.
The AAR from a short sale on day �4 followed by a short covering on day 0, referred to as R�4,0 (thenegative of the average BHAR over the [�4, 0] window) is a significant 5.66% for the financial sectorand 8.83% for the diversified financial industry. Both the financial sector and the diversified financialindustry’s AAR are significant at the 1% level. This implies that during the LBs’ bankruptcy thoseinvestors who engaged in the short-selling of financial sector and diversified financial securities overthe [�4, 0] window would have benefited. In support, Bartram and Bodnar (2009, p. 1281) state that on17 September 2008 ‘‘reports emerge that regulators are probing the practice of ‘naked’ short sellers.The SEC announces a temporary emergency ban on short selling in the stocks of all companies in thefinancial sector’’. This implies that the short selling of financial sector securities as discussed in thispaper was profitable during the LBs’ bankruptcy.
7. Conclusion
Motivated by Narayan and Sharma’s (2011) findings on sector and firm heterogeneity, this paperexamines the impact of the news announcement of the LBs’ bankruptcy on the performance of NYSEsectors and financial industries. The main contribution of this study is that it provides evidence thatsectors behave heterogeneously during a crisis and unravels that the significant adverse impact fromthe LBs’ bankruptcy is discriminatory toward the financial sector and the diversified financial industrythat were most exposed to LBs. This paper proposes for investors to short sell those sectors orindustry’s securities that are anticipated to be the most significantly adversely affected by a particularnegative news announcement.
This paper unravels three new findings during the LBs’ bankruptcy. (1) This paper providesevidence that the impact from the LBs’ bankruptcy varied across sectors and financial industries. Onthe event date and over the [�4, 0] trading days window, being the period over which most negativenews on LBs was released, not all sectors and financial industries’ performance was significantlyadversely affected. Although all financial industries were adversely affected on the event date and overthe [�4, 0] event window, not all financial industries were significantly adversely affected. (2) Thispaper finds negative news on the LBs’ bankruptcy having the greatest adverse impact on the
Table 7Findings on the average abnormal return from short selling.
Average t-Statistic
R�4,0
Financial sector 0.0566*** 3.6536
Diversified financials industry 0.0883*** 5.5483
Notes: This table reports the average abnormal return from short selling on day �4 followed by
a short covering on day 0, referred to as R�4,0 (the negative of the [�4, 0] window’s average BHAR).
t-Statistics are based on the ordinary cross-sectional approach for significance testing.*** Significance at 1% level.
K. Ranjeeni / Economic Systems 38 (2014) 178–193 191
performance of the financial sector and the diversified financial industry (both were most exposed toLBs) in comparison to other sectors and financial industries. The financial sector and the diversifiedfinancial industry’s performance were significantly adversely affected not only on the event date butalso over the [�4, 0] event window. (3) This paper shows that investors who engaged in the short saleof financial sector and diversified financial securities over the [�4, 0] window would have benefited.The AAR from a short sale on day-4 followed by a short covering on day 0 is a significant (at 1% level)5.66% for the financial sector and 8.83% for the diversified financial industry.
Finally, this study used an event study approach to analyze the performance of sectors and financialindustries during the LBs’ bankruptcy. At any particular time period, security prices are affected byvarious news announcements or events. However, under the event study approach, the abnormalperformance of firms during the event period is attributed only to the event under investigation. Thisis a limitation not only of this study but generally the limitation of the event study paradigm.Nevertheless, in event studies, it is a common practice to gather a sufficiently large sample of firmsexperiencing the event to ensure that the single commonality among the firms is the event.Consequently, other random factors are canceled and the statistical tests appropriately reflect theimpact of the event under investigation. Also, in this paper all necessary measures were taken toascertain reliable results under conditions of event-induced variances. First, the Fama and French(1993) model was used to perform analysis and the most popular MM along with the market-adjustedmodel were used to check the robustness of the results. Second, for the MM, the robust hybridBoehmer et al. (1991) standardized cross-sectional test was conducted to ensure that hypothesistesting is performed correctly.
Acknowledgments
I would like to sincerely thank Alfred Deakin Prof. Paresh Kumar Narayan, Prof. Joakim Westerlund,Dr. Huson Ali Ahmed, Dr. Kannan Thuraisamy, Dr. Sagarika Mishra and Dr. Susan Sharma from theCentre for Financial Econometrics, Deakin University for their constructive comments andoverwhelming support on earlier versions of this paper. I am also very grateful to the participantsat the internationally refereed Global Accounting, Finance and Economics Conference that was held inMelbourne in February 2012 for their useful comments. Special thanks to Dr. Rohitash Chandra (TheUniversity of the South Pacific) and Mr. Dharmendra Naidu (The University of the South Pacific) fortheir helpful feedback. I also thank God for his guidance and my family for their continuous moralsupport in the completion of this paper.
References
Acharya, V., Philippon, T., Richardson, M., Roubini, N., 2009. The financial crisis of 2007–2009: causes and remedies. Financ.Markets Inst. Instrum. 18 (2) 89–137.
Ball, R., Brown, P., 1968. An empirical evaluation of accounting income numbers. J. Acc. Res. 6 (2) 159–178.Barber, B.M., Lyon, J.D., 1997. Detecting long-run abnormal stock returns: the empirical power and specification of test statistics.
J. Finan. Econ. 43 (3) 341–372.Bartram, S.M., Bodnar, G.M., 2009. No place to hide: the global crisis in equity markets in 2008/2009. J. Int. Money Finan. 28 (8)
1246–1292.Beaver, W.H., 1968. The information content of annual earnings announcements. J. Acc. Res. 6 (3) 67–92.Boehmer, E., Masumeci, J., Poulsen, A.B., 1991. Event-study methodology under conditions of event-induced variance. J. Finan.
Econ. 30 (2) 253–272.Brown, S.J., Warner, J.B., 1980. Measuring security price performance. J. Financ. Econ. 8 (3) 205–258.Brown, S.J., Warner, J.B., 1985. Using daily stock returns: the case of event studies. J. Finan. Econ. 14 (1) 3–31.Cable, J., Holland, K., 1996. Modelling Normal Returns in Event Studies: A Model-Selection Approach and Pilot Study. In:
ABERYSTWYTH Economic Research Papers.Cable, J., Holland, K., 1999. Regression vs. non-regressive models of normal returns: implications for event studies. Econ. Lett. 64,
81–85.Calice, G., 2011. The subprime asset-backed securities market and the equity prices of large complex financial institutions. J. Int.
Finan. Markets, Inst. Money 21 (4) 585–604.Campbell, J.Y., Lo, A.W., MacKinlay, A.C., 1997. The Econometrics of Financial Markets. Princeton University Press, Princeton, NJ.Chakrabarty, B., Zhang, G., 2010. Financial contagion channels: market microstructure evidence from Lehman Brothers
Bankruptcy.Chen, T.C., Lai, M.Y., 2013. Are investors rational? Evidence on the impact of industrial framing reclassification on stock market
reaction. J. Behav. Finan. 14 (1) 1–8.
K. Ranjeeni / Economic Systems 38 (2014) 178–193192
Chong, C.Y., 2011. Effect of subprime crisis on us stock market return and volatility. Global Econ. Finan. J. 4 (1) 102–111.Coelho, L.R., Taffler, J., 2008. When is bad news really good news? The case of strategic vs. non-strategic bankruptcies. , pp. 1–50.Dumontaux, N., Pop, A., 2012. Contagion Effects in the Aftermath of Lehman’s Collapse: Measuring the Collateral Damage.
mimeo, , pp. 1–26., http://hal.archives-ouvertes.fr/docs/00/69/57/21/PDF/LEMNA_WP_201227.pdf.Eichler, S., Karmann, A., Maltritz, D., 2011. The term structure of banking crisis risk in the United States: a market data based
compound option approach. J. Banking Finance 35, 876–885.Fama, E., Fisher, L., Jensen, M., Roll, R., 1969. The adjustment of stock prices to new information. Int. Econ. Rev. 10, 1–21.Fama, E.F., French, K.R., 1993. Common risk factors in the returns on stocks and bonds. J. Financ. Econ. 33 (1) 3–56.Foster, G., 1980. Accounting policy decisions and capital market research. J. Acc. Econ. 2 (1) 29–62.Gorban, A.N., Smirnova, E.V., Tyukina, T.A., 2010. Correlations, risk and crisis: from physiology to finance. Phys. A: Stat. Mech.
Appl. 389 (16) 3193–3217.Harrington, S.E., Shrider, D.G., 2007. All events induce variance: analyzing abnormal returns when effects vary across firms. J.
Financ. Quant. Anal. 42 (01) 229–256.Kaufman, G.G., 1994. Bank contagion: a review of the theory and evidence. J. Financ. Serv. Res. 8 (2) 123–150.Kryzanowski, L., Switzer, L., Jiang, L., 1995. Stock market crash behavior of screen-sorted portfolios. Int. Rev. Econ. Finance 4 (3)
227–244.Lee, S., Connolly, D.J., 2010. The impact of IT news on hospitality firm value using cumulative abnormal returns (CARs). Int. J.
Hosp. Manage. 29 (3) 354–362.Longstaff, F.A., 2010. The subprime credit crisis and contagion in financial markets. J. Financ. Econ. 97 (3) 436–450.MacKinlay, A.C., 1997. Event studies in economics and finance. J. Econ. Lit. 35 (1) 13–39.Mio, C., Fasan, M., 2012. Does corporate social performance yield any tangible financial benefit during a crisis? An event study of
Lehman Brothers’ bankruptcy. Corp. Reput. Rev. 15 (4) 263–284.Narayan, P.K., Sharma, S.S., 2011. New evidence on oil price and firm returns. J. Banking Finance 35, 3253–3262.Patell, J.M., 1976. Corporate forecasts of earnings per share and stock price behavior: empirical tests. J. Acc. Res. 14 (2) 246–276.Patell, J.M., Wolfson, M.A., 1979. Anticipated information releases reflected in call option prices. J. Acc. Econ. 1 (2) 117–140.Peterson, P.P., 1989. Event studies: a review of issues and methodology. Quart. J. Bus. Econ. 28 (3) 36–66.Pichardo, C., Bacon, F., 2009. The Lehman Brothers bankruptcy: a test of market efficiency. Proc. Acad. Acc. Financ. Stud. 14 (1)
43–48.Raddatz, C., 2010. When the rivers run dry: liquidity and the use of wholesale funds in the transmission of the US subprime
crisis. World Bank 5203, 1–54.Ritter, J.R., 1991. The long-run performance of initial public offerings. J. Finance 46 (1) 3–27.Samarakoon, L.P., 2011. Stock market interdependence, contagion, and the US financial crisis: the case of emerging and frontier
markets. J. Int. Financ. Markets Inst. Money 21 (5) 724–742.Schwert, G.W., 1981. Using financial data to measure effects of regulation. J. law Econ. 121–158.Trueman, B., Wong, M.H.F., Zhang, X.-J., 2003. Anomalous stock returns around Internet firms’ earnings announcements. J. Acc.
Econ. 34 (1–3) 249–271.Wang, Z., Kutan, A.M., Yang, J., 2005. Information flows within and across sectors in Chinese stock markets. Quart. Rev. Econ.
Finance 45 (4/5) 767–780.
K. Ranjeeni / Economic Systems 38 (2014) 178–193 193
THE UNITED STATES STOCK MARKET PERFORMANCE
DURING LEHMAN BROTHERS BANKRUPTCY
Kumari Ranjeeni* and Rohit Kishore**
In this paper, we examine the impact of Lehman Brothers (LB) bankruptcy on the US stock market performance. The investigation is conducted for a total of 488 firms listed on NYSE using an event study approach. We find evidence in support of our propositions that LB bankruptcy most significantly influenced the banking industry and the financial sector performance
Field of Research: Global Financial Crisis _____________________
* Miss Kumari Ranjeeni, School of Accounting and Finance, Faculty of Business and Economics, The University of the South Pacific, Fiji.
** Dr. Rohit Kishore, School of Accounting and Finance, Faculty of Business and Economics, The University of the South Pacific, Fiji.
119
APPENDIX B
1. Insert letter from the editor of the journal, “Emerging Markets, Finance and Trade” stating that the paper entitled “The Impact of the Lehman Brothers’ bankruptcy on the performance of Chinese sectors” has been accepted for publication.
2. Insert the paper entitled “The impact of the Lehman Brothers’ bankruptcy on the performance of Chinese sectors”.
1
THE IMPACT OF THE LEHMAN BROTHERS’ BANKRUPTCY ON THE
PERFORMANCE OF CHINESE SECTORS
Kumari Ranjeeni1, Susan Sunila Sharma
Centre for Economics and Financial Econometrics Research, School of Accounting, Economics and Finance, Faculty of Business and Law, Deakin University, 221 Burwood Highway,
Burwood, Victoria 3125, Australia.
ABSTRACT
This paper investigates the impact of the news announcement of the Lehman Brothers’ (LBs’)
bankruptcy on the performance of Shanghai Stock Exchange (SSE) sectors. Unlike the
assumption in this literature that firms are homogenous, we address the unknown issue: does
LBs’ bankruptcy have a heterogeneous effect on stock returns of sectors listed on SSE? We find
statistically insignificant effect of LBs’ bankruptcy on the performance of energy and financial
sector while most of the other sectors suffered significantly. Thus, our results highlight on the
heterogeneous effect of LBs’ bankruptcy on different sectors and at different time intervals
surrounding the event.
KEYWORDS: Lehman Brothers’ bankruptcy, Global Financial Crisis, abnormal returns,
Chinese sectors, event study.
JEL: G01, G11, G14, G33
1 Corresponding author. Tel.: +61 410 199 805; Fax: +61 3 924 46034, E-mail address: [email protected]
2
1. INTRODUCTION
Studies have largely analyzed the impact of LBs’ bankruptcy on the performance of stock returns
in the US stock market (see for example, Dumontaux and Pop, 2009; Pichardo and Bacon, 2009;
Ranjeeni, 2014). Dumontaux and Pop (2009) examined the performance of the US surviving
financial institutions (disaggregated large banks and “non-bank” Diversified Financial Industry)
for the year 20082 and find contagion effects from LBs’ bankruptcy discriminatory towards
biggest financial firms in the financial sector with significant effects on surviving “non-bank”
financial services firms (mortgage and speciality finance, investment services and diversified
financial services firms). Pichardo and Bacon (2009) analyzed the impact of the LBs’ bankruptcy
on the performance of 15 investment firms over the period 1 September 2008 to 27 October
2008. They find the investment firms’ stock prices on and around the event date significantly
negatively affected from LBs’ bankruptcy. Ranjeeni (2014) investigated the impact of LBs’
bankruptcy on the performance of New York Stock Exchange (NYSE) sectors and financial
industries over the period 9 September 2008 to 15 September 2008. Ranjeeni (2014) documents
that the impact of LBs’ bankruptcy varied across NYSE sectors and financial industries. She
finds that the financial sector and the diversified financial industry were most significantly
adversely affected during LBs’ bankruptcy.
There is less evidence on the effect of LBs’ bankruptcy on the Chinese market. The US stock
market differs from the Chinese stock market both in terms of nature and composition. China’s
capital market is relatively closed and state controlled (Bianconi et al., 2013) that provides
opportunities for diversification (Bhar and Nikolova, 2009a, b). China’s stock index is
dominated by financial companies while the US stock market has a diversified composition
2 The investigation was done from −2 to +2 days from LBs’ bankruptcy.
3
(Bianconi et al., 2013). According to the 15 October 2013 Shanghai Stock Exchange Composite
Index constituents’, while financial companies alone constitute the majority, 41 per cent of the
total current market capitalization, each of the other nine sectors representation is less than 20
per cent3. Limited studies have examined the performance of Chinese stock returns during LBs’
bankruptcy. Raddatz (2010)4 is the only study that used an event study methodology to examine
the effect of LBs’ bankruptcy on the performance of Chinese stock returns of ten individual
banks. Raddatz (2010) revealed that globally and within countries, those banks relying heavily
on non-deposit sources of funds experienced a substantial fall in stock returns. On the other
hand, Bianconi et al. (2013) examined the impact of the US financial crisis on the performance
of Chinese stock returns at aggregate market level using daily data from January 2003 to July
2010. Bianconi et al. (2013) provides evidence that the effect of the US financial stress on the
Chinese stock market is negligible and much less relative to Brazil, Russia and India.
A limitation of Bianconi et al. (2013) market level analysis is that all the firm and sector
constituents of the market are incorrectly assumed to be homogenous in nature. A branch of
research has revealed that the share market firms and sectors are heterogeneous in nature (see for
instance, Pennings and Garcia, 2004; Beltratti, 2005; Hanson et al., 2008; Narayan and Sharma,
2011) due to the varying levels of industry concentration (see inter alia, Hou and Robinson,
2006), investors information processing ability (see Merton, 1987; Schiller, 2001; Sims, 2001;
Hong et al., 2007), speed of information flow to different industries (see for example, Narayan
and Sharma, 2011), and firm-specific characteristics, the structure of the industry to which firms
belong and the state of the economy (Chou et al., 2012). LBs’ bankruptcy occurs in the US 3 Each of the other nine Chinese sectors representation is as follows: industrials and energy (15%), materials and consumer discretionary (8%), consumer staples (4%), utilities, health care, and information technology (3%) and telecommunication services (1%). 4 Raddatz (2010) examined the impact of LBs’ bankruptcy on stock price returns of 662 individual banks across China and other 43 countries excluding the US.
4
financial sector. Since Chinese stock market is dominated by financial companies and the
Chinese financial and energy sectors are weakly integrated with the US stock market (Bhar and
Nikolova, 2009a, b; Morales and Gassie, 2011), the market level analysis of Bianconi et al.
(2013) is likely to be influenced by the financial companies performance. Therefore, a
disaggregate level analysis is needed to analyze the performance of the Chinese stock market.
The gap in the literature is the absence of disaggregate level analysis to examine the immediate
impact of LBs’ bankruptcy on the performance of Chinese sectors. We use an event study
methodology to fill this research gap. Such an investigation is essential for the following reasons.
First, disaggregate sectoral level analysis will show whether any of the Chinese sectors were
significantly affected during LBs’ bankruptcy. “Firms of different industries … may have
different sensitivities to business cycles …” (Chou et al., 2012, p.359) depending on the products
produced and the stage of the industry’s life cycle. This implies that the impact of LBs’
bankruptcy may vary across sectors. In support, Ranjeeni (2014) provides evidence that LBs’
bankruptcy had varied impact on the performance of sectors in the US stock market. Therefore,
sectoral level analysis will reveal whether Chinese sectors performed heterogeneously or
homogeneously during the volatile period of LBs’ bankruptcy announcement.
Second, our analysis will enable the determination of whether Chinese financial and energy
sectors, which are weakly integrated with the US, were insulated from the impact of LBs’
bankruptcy. This will provide insights on whether Chinese financial and energy sectors could
provide cross-country diversification opportunities for US investors. Finally, using an event
study methodology, we examine the evolution of cumulative returns at sector level from four
trading days prior to LBs’ bankruptcy till the date on which LBs’ bankruptcy is announced. The
aforementioned five trading days is dominated by negative news announcements on LBs’
5
(Bartram and Bodnar, 2009). Since the speed at which information flows to industries vary
(Narayan and Sharma, 2011) along with investors information processing ability, our event study
analysis will provide insights on the immediate response of each of the Chinese sectors during
LBs’ bankruptcy.
The balance of the paper is organized as follows. In the next section, we discuss the data and
methodology used in this study. In section 3, we provide main findings. In the final section, we
provide some concluding remarks.
2. DATA AND METHODOLOGY
2.1. Data
We use Shanghai Stock Exchange (SSE) Composite Index as the market index and examine the
performance of its sector constituents. The 15 October 2013 SSE Composite Index constituents’
firm data list, which contained information on company name, ticker symbol and Global Industry
Classification Standards (GICS) are downloaded from Bloomberg database. Following this, each
of the firms’ ticker symbols are used to download their respective last price data expressed in
USD from Bloomberg database. In order to avoid biasness in the results, we removed those firms
from the sample that had missing data during the estimation period which is from 14 August
2006 to 29 December 20065. Consequently, the sample used in this paper consists of a total of
845 firms. These firms are divided into nine sectors6 according to their respective GICS. These
nine sectors are consumer discretionary (consists of 148 firms), consumer staples (consists of 62
firms), energy (consists of 19 firms), financial (contains 92 firms), health care (contains 60
5 Specific details are available upon request. 6 We excluded Telecommunication Services sector because there were only two firms that belonged to this sector.
6
firms), industrial (contains 205 firms), information technology (contains 59 firms), material
(contains 156 firms), and utility (contains 44 firms) sectors.
2.2. Methodology
Our empirical analysis uses an event study approach7 to examine the impact of the news
announcement on LBs’ bankruptcy on the stock returns of the SSE sectors. Analogous to
Dumontaux and Pop (2009), Pichardo and Bacon (2009), Raddatz (2010), Mio and Fasan (2012)
and Ranjeeni (2014), we define the event date as the news announcement of LBs’ bankruptcy
that occurs in the US on Monday 15 September 2008. On “September 15, 2008 (Mon) ... 12.30
am EST: Lehman Brothers Holdings Incorporated files for Chapter 11 bankruptcy protection.
SEC Filing ...” (Bartram and Bodnar, 2009, pp 1280-1281). We then synchronise this event date
with the corresponding date in China. Since China is 13 hours ahead of US, the event date of 15
September 2008, 12.30 am relates to 1.30 pm in China on the same calendar day. We allow some
time for the market participants in China to become aware of LBs’ bankruptcy. The SSE trading
hours in the afternoon session for continuous auction opens at 1.00 pm and closes at 3.00 pm
(Shanghai Stock Exchange, 2010). Consequently, the event of LBs’ bankruptcy that occurred on
15 September 2008 at 12.30 am in the US will be better reflected on 16 September 2008 in
China. Accordingly, in this paper, the event of LBs’ bankruptcy corresponds to 16 September
2008 in China.
7 Event study was introduced in the late 1960s by seminal works of Ball and Brown (1968) and Fama et al. (1969). Zhu and Jog (2012), Elekdag and Wu (2013) and Hasan and Xie (2013) used an event study methodology to examine stock returns in emerging markets.
7
We choose 15 September 2008 in the US as the event date due to the following reasons. First,
despite an enormous loss of $4 billion reported by LBs’ on 10 September 20088 (see Bartram and
Bodnar, 2009), the event of 15 September 2008 stands out due to the extent and pace of its
alarming consequences (Raddatz, 2010). “September 15, 2008 has been proclaimed Wall Street’s
worst day in seven years. The Dow Jones Industrial average lost more than 500 points, more than
4%, which is the steepest fall since the day after the September 11th attacks” (Pichardo and
Bacon, 2009, p. 44). On 15 September 2008, LBs’ had total assets of $639 billion making it the
largest failure in the US history. This stimulates the need to examine the response of the Chinese
stock market to the catastrophic news on the immense failure of the largest fixed interest security
dealer, LBs’. Second, it is the news announcement of LBs’ bankruptcy on 15 September 2008
that triggered the GFC. Despite the ongoing mortgage and banking crisis since early 2007, the
bankruptcy of LBs’ on 15 September 2008 marks the “real” anticlimax era in the stock market
performance (Bartram and Bodnar, 2009). It triggered overwhelming declines in index levels,
escalated price volatility universally (Bartram and Bodnar, 2009; Chong, 2011; Eichler et al.,
2011; Samarakoon, 2011) and infected the entire global financial system (Eichengreen et al.,
2012). Third, the intense transmission of financial crisis from the US to emerging markets also
starts following the bankruptcy of LBs’ (Dooley and Hutchison, 2009; Bianconi et al., 2013) and
was recorded higher than the peak levels during the Asian crisis (Balakrishnan et al., 2011). The
emerging economies’ response at the start of the subprime crisis was limited relative to the US
and other industrial economies (Dooley and Hutchison, 2009). Finally, the findings of this paper
will reveal whether investors’ could benefit by investing in any of the Chinese sectors during
such a volatile period.
8 Our event window of [−4, 0] trading days incorporates the effect of LBs’ huge loss reported on 10 September 2008.
8
Similar to Ranjeeni (2014), we use a shorter event window of 5 trading days to control for
multiple negative news announcements around LBs’ bankruptcy impairing the results. Two such
negative news announcements were as follows. First, Fannie Mae and Freddie Mac were placed
into government conservatorship by the Federal Housing Finance Agency on Sunday, 7
September 2008 (see Bartram and Bodnar, 2009) and the subsequent day was 5 trading days
prior to LBs’ bankruptcy. Second, American International Group (AIG) was bailed out on the
day subsequent to the event date. To control for the aforementioned negative news
announcements, we exclude 5 trading days prior to LBs’ bankruptcy and the days subsequent to
the event date from the event window. According to the timeline of events provided in Bartram
and Bodnar (2009)9, negative news announcements on LBs’ starts from 4 trading days prior to
LBs’ bankruptcy (9 September 2008). Therefore, the impact of the LBs’ bankruptcy is analyzed
for an event window of [−4, 0] trading days. The [−4, 0] trading days ranges from 10 September
2008 to 16 September 2008 in China.
We measure abnormal returns using the following two models. The first model (Equation 1) is
the most commonly used Fama et al. (1969) market model10 that is based on the Capital Asset
Pricing Model (CAPM) (for related studies see Brown and Warner, 1985; Peterson, 1989;
MacKinlay, 1997; Song and Walkling, 2000; Miyajima and Yafeh, 2007; Chern et al., 2008; Lee
and Connolly, 2010; Butler et al., 2011; Chen and Lai, 2013). Similar to Raddatz (2010) and
Ranjeeni (2014), we use an estimation period prior to the GFC to forecast normal period returns
9 For a detailed timeline of events that occurred during the GFC, refer to Bartram and Bodnar (2009). 10 The Fama et al., (1969) market model is superior to the market adjusted returns model (referred to as Model 2 in this paper). This is because the market adjusted returns model is a restricted market model with alpha constrained to zero and beta constrained to be one (MacKinlay, 1997).
9
that would have eventuated in the absence of GFC11. Federal Reserve Bank of St. Loius (2014)
identifies the crisis period ranging from 27 February 2007 to 30 December 2009. However, in
January 2007, an event known as "Chinese Correction" triggered high volatility in Chinese
markets (Bianconi et al., 2013). In order to avoid the event of Chinese correction impairing our
estimation period results, we exclude the month of January 2007 and use an estimation period of
100 days before 31 December 2006. The estimation period of 100 trading days used in this paper
falls within the typical lengths of estimation period (see Peterson, 1989) and is defined as 14
August 2006 to 29 December 2006.
In the estimation period, for each firm, share price returns at time t is regressed on the concurrent
market index return. The estimated alpha and beta ( coefficients obtained from the
model’s prediction are used to calculate the expected return for each firm at time
during the event period.
(1)
where; represents expected return for firm at time ; is the estimated alpha for firm ;
is the estimated beta for firm ; is the SSE Composite Index return at time .
The expected returns are considered to be “normal returns” which are unaffected by the event.
Abnormal returns are the difference between stock’s actual realized return during
the event and the predictions based on the Fama et al. (1969) market model. 11 Since the GFC was a period characterized by high levels of volatility, it will be inappropriate to have an estimation period immediately prior to the event of LBs’ bankruptcy (see Ranjeeni, 2014).
10
(2)
where represents abnormal return for firm at time ; is the actual realized return for
firm at time ; represents expected return for firm at time .
Similar to the conventional event study reporting of cumulative returns (Raddatz, 2010), we
focus on the evolution of cumulative average abnormal returns (CAAR) over the event window
of [−4, 0] trading days12. However, the daily average abnormal returns (AAR) on the event date
are also reported. We use the standardized cross-sectional test statistic proposed by Boehmer et
al. (1991)13 to test the statistical significance of AR as it provides robust results in the presence
of both event date clustering14 and event induced variances (Boehmer et al., 1991).
The second model is the market adjusted returns model (see Brown and Warner, 1985). Brown
and Warner (1985) find that for event studies conducted using daily data, the market adjusted
return model performs reasonably well and is also powerful in the presence of event date
clustering. This model is used in scenarios whereby “it is not feasible to have a pre-event
estimation period for the normal model parameters” (MacKinlay, 1997, p.18). Accordingly, we
use the market adjusted returns model as it does not depend on past returns to forecast normal
period returns (Dumontaux and Pop, 2009) during LBs’ bankruptcy. Under the market adjusted
12 Negative news announcements on LBs’ dominated the event window of [−4, 0] trading days (Bartram and Bodnar, 2009). 13 The Boehmer et al. (1991) standardized cross-sectional test is a hybrid formed by combining the two approaches of standardized residuals (Patell, 1976) and the ordinary cross-sectional method (see Brown and Warner, 1980). 14 Since we examine the sample of securities at the sector level, cross-sectional dependence in the AR arising from event date clustering can occur to a greater extent and cause measurable misspecifications (Brown and Warner, 1985).
11
return model, the normal period return is the current event period market return. We follow
Brown and Warner (1985) and use the ordinary cross-sectional test approach to test the statistical
significance of the Model 2 AAR and CAARs.
3. MAIN FINDINGS
In this section, we discuss the event study results for nine sectors. We present the findings of
AAR and CAARs using Model 1 (CAPM) for each of the sectors in Panel A of Table 1,
respectively. Results based on AAR reveal that there is statistically significant effect of LB’s
bankruptcy on stock returns of five sectors, namely energy, healthcare, industrial, material, and
utility. The CAAR for [-4,0] trading days event window show that LBs’ bankruptcy has a
statistically significant effect on stock returns of seven sectors, namely consumer discretionary,
consumer staples, health care, industrial, information technology, materials, and utility sectors.
In the case of material sector, there is significant effect of LBs’ bankruptcy only on the event
date, over the [-4, 0] trading days window and four days before the event date. We also
document that there is statistically insignificant effect of LBs’ bankruptcy on stock returns of
energy (except on the event date) and financial sectors. In contrast, Ranjeeni (2014) documents
that the stock returns of energy and financial sectors in the US market were significantly affected
during LBs’ bankruptcy. The CAAR of the US financial sector was significantly adversely
affected (at 1% level) over all multi day intervals analyzed in this paper (Ranjeeni, 2014). Apart
from [-4,-3] and [-4,-2] multi day intervals, the CAAR of the US Energy sector was largely
significantly affected on the other intervals (Ranjeeni, 2014). This reveals that Chinese stock
market and the sectors that constitute the market behave differently compared to the same sectors
in the US market (Ranjeeni, 2014). The contradiction between the sectors of the US and Chinese
12
stock market is due to the fact that Chinese financial and energy sectors are weakly integrated
with the US stock market (Bhar and Nikolova, 2009a, b; Morales and Gassie, 2011). This
implies that Chinese energy and financial sectors can provide cross-country diversification
opportunities for US investors during volatile periods.
Moreover, Bianconi et al. (2013) also provides evidence that the effect of the US financial stress
on the Chinese stock market is negligible. However, our sector-level analysis provides evidence
in support of sector heterogeneity (Narayan and Sharma, 2011 and Ranjeeni, 2014) by showing
that some of the sectors in the Chinese stock market are statistically significantly affected while
others are statistically insignificantly affected by LBs’ bankruptcy. This implies that since
Chinese stock market is dominated by financial companies, the market level analysis by
Bianconi et al. (2013) is influenced by the performance of the financial sector.
There are two other aspects of our findings which are worth highlighting. First, the AAR results
reported in Panel A of Table 1 reveals that the LBs’ bankruptcy has a positive effect on stock
returns of four sectors (namely healthcare, industrial, material, and utility). This result is
consistent with the work of Ranjeeni (2014) where she documents that the same four sectors in
the US market are significantly positively affected during LBs’ bankruptcy. Second, in Panel A
of Table 1, we notice in the case of five sectors, namely consumer discretionary, healthcare,
industrial, information technology, and utility sectors, the LBs’ bankruptcy has a consistent
positive effect on stock returns in all five different widow combinations. In addition, in the case
of consumer staples and material sectors, LBs’ bankruptcy positively affected stock returns in
four ([-4], [-4,3], [-4,1], [-4,0]) and two ([-4], [-4,0]) out of five event windows, respectively.
These sectors are positively affected due to the fact that these sectors are stable sectors of the
economy with an inelastic demand. Consequently, even in a recession, people will continue to
13
demand for basic amenities. This reveals there is heterogeneous effect of LBs’ bankruptcy on
different sectors and at different time intervals surrounding the event. Our findings support our
motivation for conducting this analysis at the sector level.
The results based on market adjusted return model is reported in Panel B of Table 1. The AAR
based results show that LBs’ bankruptcy has a positive and statistically significant effect on the
stock returns of all eight sectors except in the case of Energy sector. The CAAR based results
reported in Panel B of Table 1 is consistent with what we earlier document using CAPM based
return model. Again, we are able to report that the stock returns of four sectors (consumer
discretionary, industrials, information technology, and utility) are consistently positively affected
by the LBs’ bankruptcy during all five trading days event window.
INSERT TABLE 1
3.1. Robustness test
As a robustness test, we check whether the Fama et al. (1969) market model (Model 1) results
are influenced by the day-of-the-week (DOW) effects using the following model adopted from
Gibbons and Hess (1981)15. For each firm, share price return at time is regressed on the
concurrent market index return and dummy variables for each DOW except Wednesday. The
estimated alpha and beta ( coefficients obtained from the model’s prediction are used to
calculate the expected return and abnormal return for each firm at time during the event period
using equations 1 and 2, respectively. The DOW effects model has the following form:
(6)
15 A similar analysis has been used by Ajayi et al., (2004) to examine DOW stock return anomaly in Eastern European Emerging Markets.
14
Where is the actual realized return for firm at time ; is the SSE Composite Index
return at time ; is a dummy variable for Monday (i.e., =1 if observation falls on
Monday and zero otherwise); is a dummy variable for Tuesday (i.e., =1 if observation
falls on Tuesday and zero otherwise); is a dummy variable for Thursday (i.e., =1 if
observation falls on Thursday and zero otherwise); is a dummy variable for Friday (i.e.,
=1 if observation falls on Friday and zero otherwise).
The results based on the DOW adjusted returns are reported in Table 2. The AAR based results
(see Panel A of Table 2) reveal that, there is statistically significant and negative effect of LBs’
bankruptcy on stock returns of energy sector. This result is consistent with CAPM based model.
However, the statistically significant and positive effect of LBs’ bankruptcy on stock returns has
reduced to only one sector, namely healthcare sector. In the case of other three sectors (namely,
industrial, material, and utility) where CAPM adjusted returns reported statistically significant
and positive results, has become insignificant when we adjusted returns using DOW dummies.
INSERT TABLE 2
Furthermore, the CAAR based results (see Panel B of Table 2) also report less cases of
statistically significant effect of LBs’ bankruptcy on stock returns. We find that only in two
cases, industrial and utility, the LBs’ bankruptcy consistently affected the stock returns during all
five event window trading days. It is also worth highlighting that when we adjusted stock returns
using DOW dummies, we find that LB’s bankruptcy has a statistically significant and negative
effect on stock returns of the financial sector on three out of five trading days event window, [-
4,2], [-4,1], and [-4,0]. In contrast, the US based results of Ranjeeni (2014) show that the
financial sector was significantly adversely affected (at 1% level) over all five event window
15
trading days. On the other hand, in the case of Chinese energy sector, there is insignificant
effects of LBs’ bankruptcy on stock returns over all five event window trading days. The
Chinese energy sector results are consistent with the CAPM based results. Overall, the Chinese
financial and energy sector results support the earlier implication on providing cross-country
diversification opportunities for US investors during volatile periods arising from weak
integration of both energy and financial sectors.
4. CONCLUSION
Based on the market level analysis, Bianconi et al. (2013) provide evidence that the effect of the
US financial stress on the Chinese stock market is negligible. Since Chinese stock market is
dominated by financial companies and the Chinese financial and energy sectors are weakly
integrated with the US stock market (Bhar and Nikolova, 2009a, b; Morales and Gassie, 2011),
the market level analysis of Bianconi et al. (2013) is likely to be influenced by the financial
companies performance. Motivated by Narayan and Sharma (2011) and Ranjeeni (2014) sectoral
level analysis, we use a disaggregate approach to examine the impact of LBs’ bankruptcy on the
performance of Chinese sectors.
The main contributions of this paper are as follows. First, our results provide evidence that
unlike in the US, LBs’ bankruptcy had insignificant effect on the performance of energy (except
on the event date) and financial sectors in the Chinese stock market during the investigated
period. This implies that energy and financial sectors in the Chinese stock market can provide
cross-country diversification opportunities for US investors during volatile periods. Second, our
16
results show that the Chinese market level analysis is influenced by the performance of the
financial sector during LBs’ bankruptcy as although the impact on the financial sector was
negligible, most sectors suffered significantly. This implies that the Chinese market level
analysis in Bianconi et al. (2013) is influenced by the performance of the financial sector.
Finally, our results highlight on the heterogeneous effect of LBs’ bankruptcy on different
Chinese sectors and at different time intervals surrounding the event.
Finally, we used an event study approach to analyze the performance of Chinese sectors during
LBs’ bankruptcy. A limitation of the event study paradigm is that the abnormal performance of
firms during the event period is attributed only to the event under investigation. However,
security prices are simultaneously affected by various news announcements or events.
Nevertheless, in line with the event study literature, we gathered a sufficiently large sample of
securities experiencing the event such that other random factors are negated and the single
commonality among the firms is the event. We also undertook all necessary measures to
ascertain reliable results under conditions of event induced variances. First, we used the most
popular Fama et al. (1969) market model along with the market adjusted returns model to do the
analysis. Second, we used the robust Boehmer et al. (1991) standardized cross-sectional test to
examine the significance of the Fama et al. (1969) market model results. Finally, we adjusted for
the day of the week effects in the Fama et al. (1969) market model to check the robustness of the
results.
17
REFERENCES
Ajayi, R. A., S. Mehdian, and M. J. Perry. 2004. The Day-of-the-Week Effect in Stock Returns.
Emerging Markets Finance and Trade 40 (4), 53-62.
Balakrishnan, R., Danninger, S., Elekdag, S., Tytell, I., 2011. The transmission of financial stress
from advanced to emerging economies. Emerging Markets Finance and Trade 47
(Supplement 2), 40-68.
Ball, R., Brown, P., 1968. An empirical evaluation of accounting income numbers. Journal of
Accounting Research 6 (2), 159-178.
Bartram, S.M., Bodnar, G.M., 2009. No place to hide: The global crisis in equity markets in
2008/2009. Journal of International Money and Finance 28 (8), 1246-1292.
Beltratti, A., 2005. Capital market equilibrium with externalities, production and heterogeneous
agents. Journal of Banking & Finance 29 (12), 3061-3073.
Bhar, R., Nikolova, B., 2009a. Oil prices and equity returns in the BRIC countries. The World
Economy 32 (7), 1036-1054.
Bhar, R., Nikolova, B., 2009b. Return, volatility spillovers and dynamic correlation in the BRIC
equity markets: An analysis using a bivariate EGARCH framework. Global Finance
Journal 19 (3), 203-218.
Bianconi, M., Yoshino, J.A., de Sousa, M.O.M., 2013. BRIC and the US Financial Crisis: An
Empirical Investigation of Stocks and Bonds Markets. Emerging Markets Review 14, 76-
109.
Boehmer, E., Masumeci, J., Poulsen, A.B., 1991. Event-study methodology under conditions of
event-induced variance. Journal of Financial Economics 30 (2), 253-272.
18
Brown, S.J., Warner, J.B., 1980. Measuring security price performance. Journal of Financial
Economics 8 (3), 205-258.
Brown, S.J., Warner, J.B., 1985. Using daily stock returns: The case of event studies. Journal of
Financial Economics 14 (1), 3-31.
Butler, A.W., Cornaggia, J., Grullon, G., Weston, J.P., 2011. Corporate financing decisions,
managerial market timing, and real investment. Journal of Financial Economics 101 (3),
666-683.
Chen, T.-C., Lai, M.-Y., 2013. Are Investors Rational? Evidence on the Impact of Industrial
Framing Reclassification on Stock Market Reaction. Journal of Behavioral Finance 14
(1), 1-8.
Chern, K.-Y., Tandon, K., Yu, S., Webb, G., 2008. The information content of stock split
announcements: Do options matter? Journal of Banking and Finance 32 (6), 930-946.
Chong, C.Y., 2011. Effect of Subprime Crisis on US Stock Market Return and Volatility. Global
Economy and Finance Journal 4 (1), 102-111
Chou, P.-H., Ho, P.-H., Ko, K.-C., 2012. Do industries matter in explaining stock returns and
asset-pricing anomalies? Journal of Banking and Finance 36 (2), 355-370.
Dooley, M., Hutchison, M., 2009. Transmission of the U.S. subprime crisis to emerging markets:
Evidence on the decoupling–recoupling hypothesis. Journal of International Money and
Finance 28 (8), 1331-1349.
Dumontaux, N., Pop, A., 2009. Contagion Effects in the Aftermath of Lehman's Collapse:
Measuring the Collateral Damage. mimeo, 1-26.
19
Eichengreen, B., Mody, A., Nedeljkovic, M., Sarno, L., 2012. How the Subprime Crisis went
global: Evidence from bank credit default swap spreads. Journal of International Money
and Finance 31 (5), 1299-1318.
Eichler, S., Karmann, A., Maltritz, D., 2011. The term structure of banking crisis risk in the
United States: A market data based compound option approach. Journal of Banking and
Finance 35 (4), 876-885.
Elekdag, S., and Y. Wu. 2013. Rapid Credit Growth in Emerging Markets: Boon or Boom-Bust?
Emerging Markets Finance and Trade 49 (5), 45-62.
Fama, E., Fisher, L., Jensen, M., Roll, R., 1969. The adjustment of stock prices to new
information. International Economic Review 10 (1), 1 - 21.
Federal Reserve Bank of St. Loius, 2014. The Financial Crisis A timeline of Events and Policy
Actions.
Gibbons, M.R., Hess, P., 1981. Day of the week effects and asset returns. Journal of Business 54
(4), 579-596.
Hanson, S.G., Pesaran, M.H., Schuermann, T., 2008. Firm heterogeneity and credit risk
diversification. Journal of Empirical Finance 15 (4), 583-612.
Hasan, I., and R. Xie. 2013. Foreign Bank Entry and Bank Corporate Governance in China.
Emerging Markets Finance and Trade 49 (2), 4-18.
Hong, H., Torous, W., Valkanov, R., 2007. Do industries lead stock markets? Journal of
Financial Economics 83 (2), 367-396.
Hou, K., Robinson, D.T., 2006. Industry concentration and average stock returns. The Journal of
Finance 61 (4), 1927-1956.
20
Lee, S., Connolly, D.J., 2010. The impact of IT news on hospitality firm value using cumulative
abnormal returns (CARs). International Journal of Hospitality Management 29 (3), 354-
362.
MacKinlay, A.C., 1997. Event studies in economics and finance. Journal of Economic Literature
35 (1), 13-39.
Merton, R.C., 1987. A simple model of capital market equilibrium with incomplete information.
Journal of Finance 42 (3), 483-510.
Mio, C., Fasan, M., 2012. Does Corporate Social Performance Yield Any Tangible Financial
Benefit During a Crisis? An Event Study of Lehman Brothers' Bankruptcy. Corporate
Reputation Review 15 (4), 263-284.
Miyajima, H., Yafeh, Y., 2007. Japan's banking crisis: An event-study perspective. Journal of
Banking and Finance 31 (9), 2866-2885.
Morales, L., Gassie, E., 2011. Structural breaks and financial volatility: Lessons from BRIC
countries, Dublin Institute of Technology, Dublin1 Esmeralda Gassie. University of
Limerick May, 1-35.
Narayan, P.K., Sharma, S.S., 2011. New evidence on oil price and firm returns. Journal of
Banking and Finance 35 (12), 3253-3262.
Patell, J. M. 1976. Corporate Forecasts of Earnings per share and stock price behavior: Empirical
tests. Journal of Accounting Research 14 (2), 246 - 276.
Pennings, J.M., Garcia, P., 2004. Hedging behavior in small and medium-sized enterprises: The
role of unobserved heterogeneity. Journal of Banking and Finance 28 (5), 951-978.
Peterson, P.P., 1989. Event studies: A review of issues and methodology. Quarterly Journal of
Business and Economics 28 (3), 36-66.
21
Pichardo, C., Bacon, F., 2009. The Lehman Brothers Bankruptcy: A test of Market Efficiency.
Proceedings of Academy of Accounting and Financial Studies 14 (1), 43 - 48.
Raddatz, C., 2010. When the rivers run dry: liquidity and the use of wholesale funds in the
transmission of the US subprime crisis: World Bank, 1-54.
Ranjeeni, K., 2014. Sectoral and Industrial Performance during a Stock Market Crisis. Economic
Systems (Forthcoming).
Samarakoon, L.P., 2011. Stock market interdependence, contagion, and the US financial crisis:
The case of emerging and frontier markets. Journal of International Financial Markets,
Institutions and Money 21 (5), 724-742.
Schiller, R., 2001. Irrational Exuberance. New York: Broadway Books.
Schweitzer, R., 1989. How do stock returns react to special events? Business Review
July/August, 17-29.
Shanghai Stock Exchange, 2010. Trading Overview.
Sims, C., 2001. Rational inattention: Princeton University
Song, M.H., Walkling, R.A., 2000. Abnormal returns to rivals of acquisition targets: A test of the
acquisition probability hypothesis'. Journal of Financial Economics 55 (2), 143-171.
Zhu, P., and V. Jog. 2012. Impact on Target Firm Risk-Return Characteristics of Domestic and
Cross-Border Mergers and Acquisitions in Emerging Markets. Emerging Markets
Finance & Trade 48 (4), 79-101.
22
Table 1: Event Study results based on Market Model and Market Adjusted Return Model
Notes: This table reports results based on the Fama et al. (1969) market model and the market adjusted return model in Panel A and Panel B respectively. The Fama et al. (1969) market model results are based on an estimation period of 100 trading days before the Global Financial Crisis (14/08/2006 to 29/12/2006). AAR represents results for 9 sectors on the event date 16 September 2008. t-statistics are reported in the parenthesis. *, ** and *** denote statistical significance at 10%, 5%, and 1% levels respectively. The Fama et al. (1969) market model significance levels are based on the Boehmer et al. (1991) standardized cross sectional approach for significance testing. The market adjusted return model significance levels are based on the ordinary cross sectional approach for significance testing. Panel A: Fama et al. (1969) market model based results
Sectors AAR -4 (-4,-3) (-4,-2) (-4,-1) (-4,0) Consumer Discretionary
-0.0022 (-0.601)
0.0124*** (6.802)
0.0149*** (4.883)
0.0126*** (3.897)
0.0143*** (4.294)
0.0121*** (2.649)
Consumer Staples
0.0048 (0.878)
0.006* (1.767)
0.0127** (2.457)
0.0092 (1.48)
0.0111* (1.843)
0.0159* (1.818)
Energy -0.0161* (-1.832)
-0.0061 (-1.398)
-0.0081 (-0.742)
-0.0113 (-0.801)
-0.0094 (-0.664)
-0.0256 (-1.268)
Financials 0.0014 (-0.171)
0.0041 (1.393)
0.0013 (0.219)
-0.0088 (-1.414)
-0.0084 (-1.33)
-0.007 (-0.97)
Health Care 0.0133*** (2.628)
0.0049*** (2.581)
0.0235*** (4.906)
0.019*** (2.985)
0.0209*** (3.199)
0.0342*** (3.491)
Industrials 0.0068*** (2.894)
0.0117*** (9.157)
0.0186*** (8.159)
0.0156*** (6.651)
0.0173*** (7.284)
0.0241*** (6.266)
Information Technology
0.0052 (0.476)
0.0141*** (5.553)
0.0173*** (4.068)
0.0136*** (3.037)
0.0156*** (3.433)
0.0208** (2.507)
Materials 0.004* (1.762)
0.0032** (2.029)
-0.0004 (0.401)
0.0026 (1.147)
0.0041 (1.551)
0.0081** (2.139)
Utilities 0.0087* (1.647)
0.0105*** (5.728)
0.0175*** (3.745)
0.0359*** (5.425)
0.0377*** (5.591)
0.0464*** (4.599)
Panel B: Market Adjusted Returns based results
Sectors AAR -4 (-4,-3) (-4,-2) (-4,-1) (-4,0) Consumer Discretionary
0.0105*** (3.798)
0.0099*** (5.262)
0.0219*** (7.401)
0.0176*** (5.086)
0.0176*** (5.086)
0.0282*** (6.051)
Consumer Staples
0.0119*** (3.001)
0.0036 (1.125)
0.0154*** (3.057)
0.0097* (1.786)
0.0097* (1.786)
0.0216*** (2.924)
Energy -0.0036 (-0.428)
-0.0087* (-1.845)
-0.0015 (-0.166)
-0.0069 (-0.539)
-0.0069 (-0.539)
-0.0105 (-0.582)
Financials 0.0148*** (3.417)
0.003 (1.075)
0.0104** (2.081)
-0.0003 (-0.046)
-0.0003 (-0.046)
0.0145* (1.837)
Health Care 0.028*** (6.871)
0.0022 (1.192)
0.0317*** (7.448)
0.0251*** (3.919)
0.0251*** (3.919)
0.0531*** (5.934)
Industrials 0.0158*** (6.772)
0.0094*** (7.298)
0.0229*** (10.177)
0.0179*** (7.465)
0.0179*** (7.465)
0.0337*** (8.964)
Information Technology
0.0126*** (2.578)
0.0116*** (4.258)
0.0201*** (4.668)
0.0142*** (2.850)
0.0142*** (2.850)
0.0268*** (3.499)
Materials 0.0182*** (7.438)
0.0009 (0.541)
0.0079*** (2.899)
0.0093** (2.444)
0.0093** (2.444)
0.0275*** (5.999)
Utilities 0.0223*** (4.137)
0.0079*** (4.123)
0.025*** (5.619)
0.0414*** (6.851)
0.0414*** (6.851)
0.0637*** (7.525)
23
Table 2: Robustness test – Day-of-the-week effect based results
Notes: This table reports results that has been adjusted for the DOW effects in the Fama et al. (1969) market model. The estimation period used is 100 trading days before the Global Financial Crisis (14/08/2006 to 29/12/2006). In panel A, we report AAR based results for 9 sectors on the event date 16 September 2008 and in Panel B we report results based on CAAR using event window of [-4,0] trading days. t-statistics are reported in the parenthesis. *, ** and *** denote statistical significance at 10%, 5%, and 1% levels respectively based on the ordinary cross sectional approach for significance testing.
Panel A: AAR based results Panel B: CAAR based results
-4 (-4,-3) (-4,-2) (-4,-1) (-4,0) Consumer Discretionary
-0.0047 (-1.5188)
0.01*** (5.2413)
0.01*** (2.957)
0.0054 (1.3573)
0.0047 (1.1612)
0 (0.0021)
Consumer Staples
0.0012 (0.2735)
0.0032 (1.0314)
0.0065 (1.3625)
0.0001 (0.0211)
-0.0008 (-0.1422)
0.0004 (0.0506)
Energy -0.0173* (-1.9507)
-0.0069 (-1.4513)
-0.01 (-1.107)
-0.0141 (-1.0712)
-0.0131 (-0.9792)
-0.0304 (-1.6049)
Financials -0.0005 (-0.0988)
0.0021 (0.7493)
-0.0026 (-0.5233)
-0.0147** (-2.4978)
-0.0163*** (-2.7026)
-0.0168* (-1.9027)
Health Care 0.0101** (2.1769)
0.0019 (0.9691)
0.0173*** (3.6439)
0.0098 (1.3768)
0.0086 (1.1696)
0.0187* (1.7876)
Industrials 0.0036 (1.4276)
0.0089*** (6.6281)
0.0126*** (5.0482)
0.0067** (2.4793)
0.0057** (1.9865)
0.0092** (2.0994)
Information Technology
-0.0009 (-0.1832)
0.0094*** (3.4306)
0.0068 (1.4443)
-0.0017 (-0.3088)
-0.0045 (-0.778)
-0.0054 (-0.6177)
Materials 0.0003 (0.0991)
-0.0002 (-0.1159)
-0.0075** (-2.5323)
-0.0078* (-1.9354)
-0.0097** (-2.3476)
-0.0094* (-1.8036)
Utilities 0.0072 (1.2135)
0.0095*** (4.9841)
0.0152*** (2.8991)
0.0326*** (4.9466)
0.0334*** (5.0597)
0.0407*** (4.1336)