BAILOUT DIFFERENCES

(1)

By:

Hendrik Hoekstra

University of Groningen MSc Finance

Under guidance of:

Dr. Michalis Azouras

BAILOUT DIFFERENCES

(2)

1 University of Groningen MSc Finance By: Hendrik Hoekstra - S2601036 h.hoekstra.13@student.rug.nl

Under guidance of:

Dr. Michalis Azouras

January 2017

Abstract:

On 14 October 2008, the United States of America initiated the Troubled Asset Relief Program (TARP) in response to the so-called credit crisis. This paper

examines the stock market response to two events surrounding the TARP program: the initiation of TARP and the capital injection under TARP. It then asks whether there is any difference in stock market reactions across the different U.S. states. The stock market reactions are analysed through event studies and the use of dummy variables, for the event study the market model was used. This paper found a significant positive stock market reaction to the initiation of TARP, but a significant negative stock market reaction to the capital injections, this is in correspondence with the literature surrounding TARP. This paper also found a difference in significance dependent on which market proxy was used, for the market model, something which is not discussed in other literature surrounding TARP. This paper found evidence for state effects as some of the state dummy variables were significant, and there was quite some difference in the values of coefficients. This thesis concludes that there is indeed a difference in stock market reactions towards the two TARP events, across different states.

BAILOUT DIFFERENCES

(3)

2

1. INTRODUCTION

According to Veronesi and Zingales (2010), the bailouts of 2008 are the largest interventions in the financial markets that have ever occurred. After a period of eight years, a great deal of research concerning these interventions has been done. This episode in economic history has been well documented, but there are still things that garner further investigation. This thesis will examine the effectiveness of The troubled Asset Relief Program across different states, measured through stock returns. The research question of this thesis is: “Were there any differences regarding the effectiveness of the Troubled Asset Relief Program across states in the United States of America based on stock returns?”.

The reason for possible differences existing from state to state in the United States of Amerika, hereafter named USA, is the result of a couple of reasons. The first is that banking legislation differs by state in the USA, as banks in the USA are regulated on both a state and federal level. The second is that economic conditions in the USA differ across states, such as having a higher employment rate or a higher GDP, these difference in economic conditions, might result in using the bailout money more effectively in some states as compared to other states.

First, the stock market response to the initiation of TARP was measured, followed by the stock market response though capital injections. The state effects where measured through the use of dummy state variables, using the stock market responses as dependent variable. The main findings indicate that there is evidence for state effects in the USA, and this evidence is strongest in the case of capital injections under TARP. In addition, the abnormal returns calculated through the market model differed in significance based on which market proxy was used. Although this is not an issue that is highly discussed in the event study literature surrounding TARP, it seems too important to ignore.

This thesis begins with a literature review that provides an overview of both the short-term and long-term effects of the TARP program on the USA economy. The short-term effects, as measured by an increase in stock returns, are clear. The initiation of bailout programs like TARP are met with a positive return in stock prices, while specific announcements

(4)

3

been used more frequently in the existing event-study literature surrounding TARP, than the Fama-French model.

In order to use the market model, a market proxy is required to estimate parameters that are needed to calculate the predicted return in the event window. As a market proxy for the market model and as proxy for the banking industry, the S&P 1500 and the NASDAQ Bank composite were chosen. The S&P 1500 was chosen because it seems to represent the USA economy the best, as it contains small-cap, mid-cap and big-cap firms. The NASDAQ Bank composite was chosen because it contains banks that provide all services, such as retail banking, loans, and money transmissions, and furthermore contains all the banks of the NASDAQ, which ensures that is representative of the banking industry.In the event study literature surrounding TARP, the usage of different market proxies in the market model, according to my view, has not been checked very thoroughly . The abnormal return were estimated by subtracting the predicted return from the real (stock) returns in the event window. An event window is comprised of any number of days surrounding the day on which an event occurred. Calculation of the abnormal returns is explained more thoroughly in the methodology section of this thesis. The methodology section begins by explaining the TARP program. The information concerning TARP comes from the website of the treasury of the USA. As this thesis is about the banking industry, the main focus will be on the banking programs, and specifically the capital purchase program will be examined, as this contains most, if not all of the banking bailouts.

The data from the banks was collected from the Projects.publica.org website. This website contains data on the capital injection date and the U.S. state that a particular bank operates in. The stock returns needed were found in DataStream. In total, 214 bank stocks were used. The results of this thesis are split in three parts: the TARP announcement, the capital

(5)

4

2. LITERATURE SURVEY

In this section, I provide some general literature surrounding the TARP program to make it clear where this thesis fits with regard the rest of the literature on this subject, as well as to help the reader better understand the effects of TARP. The first section discusses why bailouts may have been necessary, because if these were not needed, then perhaps TARP should not have been enacted at all. The next section describes the effects after the bailouts, and in this section I also theorise about how these effects could have led to a significant positive stock reaction. Finally, I will discuss the market reactions to different events, focusing on TARP and other rescue packages.

2.1. WHY THERE MIGHT BE A NEED FOR BAILOUTS

Before I elaborate on why a bailout event might be positive for the equity price of a bank, I will explain why governments should consider bailing out a bank in the first place. According to financial literature on the bailouts of banks, the problem with letting a bank fail is that spill-over effects occur that damage the entire financial system. For example, Goldsmith-Pinkham and Yorulmazer (2010) concluded that the bank run on Northern Rock had spill-over effects that spread to the rest of the banking system. In addition to this, Schoenmaker notes that there is a contagion effect with bank failures, and due to this intervention by authorities is warranted (Schoenmaker (1996)). Another study conducted by Sieczka, Sornette, and Holyst, (2011) noted that the bankruptcy of Lehman Brothers resulted in a significant decline in other banking activities. Many other studies have shown that there are spill-over effects from banking to the wider economy. For instance, Weiß (2012) notes that if federal governments had not intervened during, the subprime crisis in Germany and the Japanese banking crisis, a joint collapse of stocks would surely have followed these crises. Therefore, in order to fend off spill-over or contagion effects and prevent a banking crisis or keep an existing crisis from affecting the entire economy, governments bail out troubled banks.

2.2. EFFECTS AFTER THE BAILOUTS

In this section, I discuss the positive and negative effects of a bank bailout for a bank. First, I discuss the events that, if they were expected, might lead to a positive reaction in equity price. Then, I discuss the events that might result in a negative reaction.

2.2.1. POSITIVE

(6)

5

Berger, Roman, and Sedunov (2016) have shown that systematic risk after the TARP program has decreased, indicating that TARP was successful in addressing this issue. If investors knew beforehand that bailouts would result in a decrease in systematic risk, a positive effect on the stock prices would be expected. However, if they did not know, an increase in the stock prices would not be expected.

Since investors might expect a lower systematic risk as a result of a bailout, the reaction in the stock prices might be positive. Another study conducted by Berger and Roman (2015) showed that the banks that received TARP money attained competitive advantages, making them more competitive than banks that did not receive TARP money. Due to this, a bank’s equity price might increase, as a bank that is more competitive is likely more lucrative. Blau, Brough, and Thomas (2013) note that the more a bank or firm spent lobbying in the years before TARP was initiated, the more TARP money they received, meaning the TARP money can be seen as an investment paying off and the cost of lobbying and the profits from

lobbying might all be included in the stock price. Duchin and Sosyura (2014) show that banks that received TARP money appear safer if one considers the capitalisation requirements. If that effect was expected after a bailout, it would likely result in a rise in the equity price.

2.2.2. NEGATIVE

Some argue that moral hazard is encouraged through the TARP system (Black and

Hazelwood(2012)). Black and Hazelwood (2012) provide evidence that banks that received TARP money originated riskier loans. If investors expect a higher level of riskiness, they might demand a higher return, thus resulting in a negative reaction in the equity price of a bank receiving a bailout.

2.3. EVENT STUDY

This thesis uses an event study similar to the one outlined in Mackinlay (1997), which to my knowledge is one of the best articles on the use of event-study methodology. The usefulness of an event study, according to Mackinlay (1997), comes from the fact that given there is rationality in the market, an event is recorded in stock prices immediately.

2.4. MARKET REACTIONS TO CAPITAL INJECTIONS AND GENERAL ANNOUNCEMENTES

To my knowledge there is a great deal of research that uses event studies in the case of TARP. These studies have similar event dates to those used in this thesis. The most

(7)

6

All the papers that investigated the general announcements under TARP have found a positive significant stock price reaction, including: Ncube(2016); Fratianni and Marchionne (2010); Fratianni and Marchionne (2013); and Bayazitova, Dinara, and Anil Shivdasani (2012). Although Fratianni and Marchionne (2010), Fratianni and Marchionne (2013), found a

positive reaction to general announcements by country, King (2009) found differences in reactions according to country, which varied from positive to negative. However, King’s sample size is smaller than those of the afore mentioned papers, and on average the effect was found to be positive. These studies focused on countries such as the Netherlands, the United Kingdom, and the USA as a whole. In comparison, this thesis focuses on the USA only, and specifically on the differences across different U.S. states.

From the perspective of general announcements it is clear that that the market favoured the initiation of TARP. Despite this, papers that focused either on specific announcements or capital injections found a negative significant stock price reaction, including: Fratianni, and Marchionne(2010); Fratianni, and Marchionne(2013); Farruggio, Michalak, and Uhde (2013); and Elyasiani E, Mester LJ, Pagano MS. (2014). Again, King (2009) found differences

according to country1 in stock market reactions to capital injections, however, as noted above, this study has certain weaknesses.

In general, the model used for calculating the abnormal return is the market model, as seen in: King (2009); Bayazitova, Dinara, and Anil Shivdasani (2012); Ncube (2016); Farruggio, Michalak, and Uhde (2013); Joines (2010); and Elyasiani, Mester, and Pagano MS. (2014). Alternatively, an abbreviation of the market model was used, as seen in Fratianni, and Marchionne (2010). The formula concerning the market model is given below:

Rit= αi + βmt+ εit (1)

Where β is estimated as a coefficient with the market return, αi is the intercept of as stock

with the market return, and εitis the error term.

Other methods used are the Fama-French model, seen in: Ncube (2016); and Elyasiani, Mester, and Pagano MS. (2014). This model makes use of two variables, the size factor and the value factor. The model is not used by many papers and does not lead to significantly different results compared to the use of the market model.

2.4.1. SUMMING UP

From the results of these studies it is safe to say that TARP was desired by the market, as all of this research shows a positive investor reaction to the announcement of TARP. However, specific announcements and capital injections appear to lead to negative results, as

1

(8)

7

measured by the negative significant stock market reactions to capital injections and specific announcements. On this page, Table 1 is presented that compares and summarises these studies in terms of sample size, methodology, event date, stock effect, and whether or not the main objective was an event study.

Table 1:

Authors Sample Size Methodology Event Date Main Objective - Event Study Stock Market Reaction TARP or Country Data King (2009) 52 banks Stock price

Cds rates Announcements of rescue package Capital injection Yes Positive Differs by Country Country analysis Fratianni and Marchionne (2010) 120 banks for the first event. 43 for the second event.

Equity price Announcement of rescue package

Announcement that a specific bank will receive a stimulus No, part of a larger study Positive Negative Country analysis Joines (2010)

626 Equity price Announcement of TARP

Yes Positive TARP

Bayazitova, Dinara, and Anil Shivdasani (2012)

319 banks Equity price Announcement of rescue under the CPP program

Announcement of rescue under the SCAP Program Repayment of TARP Yes Positive Positive Indifferent TARP Fratianni and Marchionne (2013)

137 banks Equity price Announcement of rescue package

Announcement that a specific bank will receive a stimulus Yes Positive Negative Country analysis Farruggio, Michalak, and Uhde (2013)

707 banks Equity price Announcement

Announcement of revision

(9)

8

Elyasiani E, Mester LJ, Pagano MS. (2014)

125 Equity price Announcement of capital injections No, part of a larger study Positive TARP Ncube (2016)

214 Equity price Announcement of TARP

Capital injection under TARP

Yes Positive

Negative

TARP

(10)

9

3. DATA SECTION AND METHODOLOGY

This chapter explains the various programs that were part of TARP. This is necessary in order to provide a good understanding of TARP as this thesis investigates capital injections under the mechanisms of TARP and the announcement of TARP.

3.1. THE TARP PROGRAM

The main vehicle used by the government in response to the 2008 crisis was TARP. Initially, the U.S. congress wanted to spend $700 billion on different programs under TARP, but this number was capped at $475 billion by the Dodd-Frank Wall Street Reform and Consumer Protection Act (Dodd-Frank Act). Of this $475 billion, around $420 billion was distributed and $245 billion was invested in the banking system. After October 3, 2010 new commitments under the TARP program were no longer allowed.

Under TARP, this money was spread across five business lines: $245 billion was spent on programs to stabilize banking institutions, $27 billion was used to restart credit markets, $80 billion was used to salvage the U.S. auto industry, $68 billion was used to rescue American International Group (AIG), and $46 billion was used for programs to help struggling families avoid foreclosure. TARP included six basic programs, the auto industry, credit market programs, investment in AIG, bank investment programs, and housing and executive compensation, which are further elaborated on in subsequent sections of this thesis. The auto industry program mainly provided capital in the form of loans to major car dealers, which was necessary because the credit market was frozen. The credit market program consisted of three sub-programs: the public private investment program, the BA 7(a) securities purchase program, and the term asset backed loan facility. Housing consisted of two programs, the goal of “making homes affordable”, which gave mortgage relief to homeowners to prevent avoidable foreclosures, and the “hardest hit fund”, which provided aid to families most affected by the economic and housing market downturn. The CEO compensation program, was simply a measure that enforced that all firms that received TARP money had to have restrictions on CEO compensation, until the TARP money was repaid. The next section provides more detail regarding the banking program aspects of TARP, as these are the most relevant to this thesis.

Banking Programs

The banking investment program consisted of five key areas: the Asset Guarantee Program, the Supervisory Capital Assessment Program (SCAP) and the Capital Assistance Program (CAP), the Community Development Capital Initiative (CDCI), the Targeted Investment Program (TIP), and the Capital Purchase Program (CPP).

(11)

10

day of the agreement, and as a result termination fee of $475 million was paid by Bank of America to the USA. The agreement with Citigroup resulted in $7.1 billion in preferred shares for the treasury of the USA and in warrants to buy 66.5 million shares.

The Supervisory Capital Assessment Program was a program to assess whether banks had sufficient buffers to begin lending again. Of the participating banks, 18 of 19 had adequate capital buffers.

The Community Development Capital Initiative Program was intended to help small

communities, in these communities the traditional banks did not serve or underserve these communities. This was done through so-called “Community Development Financial

Institutions”, which provided capital mainly to low-income and middle-income individuals. The Targeted Investment Program was aimed at institutions that were considered

systematically significant. Under this program, higher rates were charged than under the CPP program. Again, only two institutions applied for this program: American Bank and Citigroup. The Capital Purchase Program had the purpose of stabilizing the financial system throughout the nation by providing capital to financial institutions. Under this program, a total of $205 billion was distributed over 707 financial institutions. Of these 707 financial institutions, 450 were small community banks. The money given was granted for preferred stock or debt securities in exchange for the investment. Most of these institutions agreed to pay a five percent dividend on preferred shares for the first five years and nine percent dividend after that. This is also the program that was investigated during the event study in this thesis, as most if not all bailouts were executed under this program.

3.2. METHODOLOGY

This section presents the methodology used to investigate whether capital injections and the general announcement of TARP led to a significant return. The event-study methodology is presented first.

3.2.1. EVENT-STUDY METHODOLOGY

The purpose of an event study is to calculate the abnormal return. The abnormal return is the return that is considered above the return that would exist under normal market circumstances (with no significant event occurring).In order to derive the abnormal return,

the expected returns are calculated, and these are subtracted from the observed or real returns, resulting in the abnormal returns.

In order to determine the expected returns and real returns in an event study, an estimation window and event window must be used. The event window is the number of days

(12)

11 Figure 1:

(Estimation window) (Event window) (Estimation window)

T0 T1 T2 T3

In order to estimate the predicted return, the market model is used (Equation 1, page 5). The market model is Rit= αi + βmt+ εit, where βmt is the coefficient of the stock return with the

stock market return, with the subscript MT being the market return, and αi is the intercept of

the stock with the market. Rit stands for the stock return of the stock in the estimation

window and εit stands for the error term. βmt and αi are estimated in the estimation window

with an Ordinary Least Squares regression, using the stock return as the dependent variable and the market return as the independent return. For the market return, the S&P 1500 and the NASDAQ Bank composite are used. The S&P 1500 is used because it contains small-cap, mid-cap and big-cap firms, while for instance an index like the S&P 500 which only contains big-cap firms, as such the S&P 1500 is seen as a better proxy for the U.S, than most indexes. economy. The NASDAQ Bank composite was chosen because it contains banks that provide all services, such as retail banking, loans, and money transmissions, and furthermore contains all the banks of the NASDAQ, which ensures that is representative of the banking industry.

These parameters are used to estimate the predicted return in the event window. For example, if the market return in the event window on day T is 5%, and βis 2andαi is 1%,

then the predicted return on that day is 5% X 2 + 1% = 11%. If the market return on the next day, day T+1, is 10%, then the predicted return on that day becomes 10% X 2 +1% = 21%. This predicted return is then used to calculate the abnormal return, the formula for which is specified below:

RAR= Rit -αi + βmt+ εit (2)

where RAR stands for the abnormal return. This abnormal return is calculated using the

(real)stock return (Rit in the event window (Rit)), and subtracting the predicted return (as

explained above) from this stock return (Rit). If the event window contains more days than just the event day, the abnormal returns are summed, creating a cumulative abnormal return. The S&P 1500 and NASDAQ Bank composite will be used for the subscript MT in equation (2).

(13)

12

operates in New York, then the dummy variable of New York gets a value of one and zero otherwise. The dependent variables are the abnormal returns.

3.2.2. EVENT WINDOW AND ESTIMATION WINDOW

As previously explained, in order to calculate the abnormal return an estimation window and an event window are needed. For the estimation window, I use the same period as Ncube (2016). For this study, this is the period from September 17, 2007 to September 17, 2008. As specified above, the estimation window is used to estimate the predicted return. Ncube (2016) considers this the normal period2, I agree with his reasoning, since in my view the first “TARP announcement” was on September 19, 2008. The event windows range from 11 days to 1 day, meaning that there is a 10 day event window, an 8 day event window, a 6 day event window, etc., specified as: (-5,+5),(-4,+4),(-3,+3),(-2,+2),(-1,+1),(0,0).

3.2.3. HYPOTHESIS

Since I expect the general announcement to be positive and the capital injection negative, and since I expect state effects to be significant for both events, the following hypotheses were generated:

General Announcement:

H0 :Abnormal return(RAR) = 0 H1: Abnormal return(RAR) > 0 Capital Injections:

H0 :Abnormal return(RAR) = 0 H1: Abnormal return(RAR) < 0 State (US) Effects of General Announcement:

H0: Dummy variable = 0 H1: Dummy variable 0

State (US) Effects of Specific Announcement:

H0: Dummy variable = 0 H1: Dummy variable 0

In order to test the significance of the abnormal returns, I use a T-test; the average of the abnormal returns is divided by the standard deviation of the abnormal returns, divided by the number of observations.

As a robustness check and since non-normality might be present, I also use the Corrado test:

2

(14)

13

( )

(3)

3.3. SAMPLE

The list of which banks were bailed out under the TARP program comes from

Projects.publica.org3. This website also contains data on which date each particular bank was bailed out, and in which state each bank operates. Projects.publica.org, a non-profit

journalism website, has the goal of providing journalism in the public interest. This

institution was founded by former managing editor of the Wall Street Journal, Paul Steiger. It has received several prizes and praises for its high standard of journalism.4

As noted in the previous section, this thesis focuses on the CPP. Although it might be more logical to obtain information from the government’s treasury website, the data concerning the bailouts was considered less complete than that of Projects.publica.org. For example, the government website does not report stock quotes or how much bailout money is received by a bank, nor does it mention in which state each bank operates. Information on the banks from the website of Projects.publica.org was collected until the announcement date of 09-01-2009. S&P 1500 returns, NASDAQ Bank composite returns, and stock returns come from the DataStream database. Companies on which DataStream does not provide any information or companies that have no trading returns for the entire estimation window were removed from the sample. The list of companies that were used can be found in Appendix 1.

Descriptive Summary

In the sample there are 214 stocks for the TARP announcement and the capital injections. The capital injections sample features 11 event dates5. In the same manner as the study by Ncube (2016), the stocks of these different event dates are grouped together. The average returns for the stocks, NASDAQ Bank composite returns and the S&P 1500, during the estimation window, as well as the average returns during the event window (-5,+5) and during the event window (0,0) for both the TARP announcement and the capital injection are shown below in Table 2.

3 https://projects.propublica.org/bailout/programs/1-capital-purchase-program visited on 16-11-2016 4 http://www.theatlantic.com/national/archive/2012/05/in-praise-of-propublica/257514/ visited on 29-11-2016 5

(15)

14 Table 2:

Returns Avg return “September

17, 2007 until September 17, 2008” (estimation window) Avg.. Return capital event window (-5,+5) Avg., return general event window (-5,+5) AVG. Return capital event day (0,0) AVG. Return general event day (0,0) NASDAQ Bank -0.07% -0.18% -0.37% 1.77% 2.95% S&P 1500 -0.08% -0.08% -0.77% 1.44% -0.73% Stock Returns -0.10% -0.29% -0.30% 0.29% 3.97%

As can be seen, the average returns in the estimation window are close (-0.07%, -0.08%, -0.10%), while the average returns in the capital event window (-5,+5) are more negative than returns in the estimation window, except for the S&P 1500. The average return in the general announcement window (-5,+5) is most negative for the S&P 1500. The correlation between banking returns and stock returns in the estimation window (not shown in the table) is 0.97, while the correlation between stock returns and the S&P 1500 returns is 0.78 This is to be expected, since an index that only features banks would probably follow the (bank) stock returns more closely than a general index. Below, two graphs show the stock returns, S&P 1500 returns, and the NASDAQ Bank composite returns during the event

window of the general announcement and the capital injections. The correlation of the stock with the S&P 1500 and NASDAQ Bank composite during the event window for the general announcement and the capital injection are: 0.79, 0.97 and 0.78, 0.85.

Graph 1: -15,00% -10,00% -5,00% 0,00% 5,00% 10,00% 15,00% -5 -4 -3 -2 -1 0 1 2 3 4 5

General announcement

(16)

15 Graph 2:

An interesting aspect surrounding the general announcement is that returns from the stock and NASDAQ Bank Composite follow each other more closely from day -1 until day 1. During the capital injections it can be seen that the S&P 1500 and the NASDAQ Bank Composite follow each other more closely than the banking stocks follow the S&P 1500 or the NASDAQ Bank Composite. Thus, it is to be expected that the differences between the S&P 1500 and the NASDAQ Bank composite will be more pronounced in the results section.

-4,00% -3,00% -2,00% -1,00% 0,00% 1,00% 2,00% 3,00% -5 -4 -3 -2 -1 0 1 2 3 4 5

Capital injections

(17)

16

4. RESULTS

In this section I provide the results of the event study and elaborate on these results, starting with the general announcement, and moving on to capital injections and the state effects.

4.1. ANNOUNCEMENT OF TARP

The sample in this study consists of 214 bank stocks. Originally, it contained 252 stocks, but due to delisting or a lack of information coverage by DataStream some stocks had to be removed. The means of the abnormal returns with their median, standard deviation, and T-value per event window using the S&P 1500 and the NASDAQ Bank are presented below. The table begins with the biggest event window and moves to the smallest:

Table 3:

S&P 1500 Event Window

Mean Median Standard Deviation T-statistic -5, +5 4.15%*** 3.00% _13.29% 4.6 -4, +4 2.93%*** 2.89% _12.60% 3.4 -3, +3 7.57%*** 8.08% _12.12% 9.1 -2, +2 9.62%*** 7.24% _15.32% 9.2 -1, +1 4.89%*** 1.59% _12.68% 5.6 -0, +0 4.61%*** 1.80% _10.25% 6.6

Notes: *,**,*** represent significance at the 10%, 5% and 1% level. Table 4:

Banking Event Window

(18)

17

As a robustness check this sample has been filtered using winsorizing at 95%6

, the results are shown in robustness section 4.1.A. This section also contains the normality results of the normal sample and the filtered sample. As can be seen above in Table 3 and Table 4, the (-5,+5) event window has less significance than the (-3,+3) event window, and the cumulative average returns are lower in the (-5,+5) window. This might indicate that in the (-5,+5) windows (Table 3 and 4) there are more negative abnormal returns than positive abnormal returns. The results in Table 3 are comparable to the results of Ncube (2016), which used a market index as opposed to an banking industry index, as the results of that paper also show a higher cumulative abnormal return in the 3,+3) window, than in the 0,+0), 1+1) and (-5,+5) windows. Also, the significance differs between what type of market proxy is used. As discussed in the literature review a positive effect was to be expected.

The first step, to check if more negative abnormal returns are present in the (-5,+5) window than in the (-3,+3) window and to see the differences between the use of the S&P 1500 and the NADSDAQ Bank Composite clearly, was to plot the daily abnormal returns in a graph during the event window. This very clearly illustrates the difference in results based on using the S&P 1500 versus the NASDAQ Bank Composite and also illustrates that the negative abnormal returns are on day -5, day -4 and day +2 till day +4. The graph shows that if the S&P 1500 is used positive returns are on day 1 and day 2, and day -1 while if the NASDAQ Bank Composite is used most of the positive returns are on day -1 and day 0.

Graph 3:

6

(19)

18

A logical explanation for the data in the graph above is that the market as a whole reacted more favourably to the stimulus-package than the banking sector alone. This explains why using the banking proxy results in lower positive returns as compared to using the banking proxy. It might also be that the banking industry more or less expected a bailout program, while the market, as a whole, perhaps did not. This might indicate why the abnormal returns for the S&P 1500 are larger than for the banking index. The abnormal returns on day -2, 0 and day 1 are quite similar to the results of the point estimation table in the paper of Ncube (2016).

The second step involved using windows that capture only the activity before the event and after the event. The post-event window features the windows (0,+5) to (0,+1), while the prior-event window features the windows (-5,0) to ( -1,0). The window (0,+5) indicates the summed average of abnormal returns for five days after and on the event date, whereas the window (-5,0) indicates the opposite (the abnormal returns summed together for five days prior to the event date and on the event date). The results are given below:

Prior-Event

Table 5:

Notes: *,**,*** represent significance at the 10%, 5% and 1% level.

Table 6:

Mean Median Standard Deviation T-statistic -5,0 3.32%*** 2.08% 12.90% 3.77 -4,0 3.03%*** 1.74% 12.45% 3.56 -3,0 4.18%*** 3.06% 11.36% 5.39 -2,0 5.57%*** 3.62% 14.26% 5.71 -1,0 0.75% -0.47% 14.63% 0.75 Banking Event Window

(20)

19

Post-event:

Table 7:

The results in the prior-event windows (-5,0 to -1,0) are clearly lower than the post-event windows (0,+5 to 0,+1). Again, the significance and the abnormal returns are more

pronounced when the S&P 1500 is used as a market proxy as opposed to when the NASDAQ Bank Composite is used. Judging from the table and the graph, it can be said that the

majority of the positive returns occurred in the days 0, 1, and 2. In both cases (Banking and S&P 1500) significance drops after event window (0,+3). Ncube (2016) found evidence that most of the abnormal returns happen in the 3 days after the event. To my knowledge, other studies in the event study literature surrounding TARP did not test research results by using an industry index and a market index to determine whether there were any meaningful differences. While King (2009) does use a banking sub-index, his sample only includes seven U.S. banks and as thus is not considered very reliable, and while Fratianni and Marchionne (2010) use an industry index that they constructed themselves, they do not check if their results are affected by using a different proxy for the market return.

Mean Median Standard Deviation T-statistic 0, +5 5.44%*** _4.18% _13.10% 5.8 0, +4 4.51%*** _2.89% _12.12% 4.8 0, +3 8.00%*** _6.20% _12.62% 8.5 0, +2 8.66%*** _7.25% _13.02% 9.2 0, +1 8.75%*** _6.94% _13.71% 9.3 Banking Event Window

(21)

20 4.2. CAPITAL INJECTION

Concerning the capital injections, the results are presented below. The same number of stocks has been used as with the general announcement.

Table 10:

All of the event windows exhibit a negative statistical significant result. The results have the highest significance on the day of the event and in the (-2+2) event window. The results in Table 10 are similar to those of Ncube (2016) where the (cumulative) abnormal returns in the (-5,+5) window are more negative than the (-3,+3) window, which is more negative than the (-1,+1) window. The negative abnormal return is the highest in the (-5,+5) S&P 1500 event window. As a robustness check, the samples have been filtered using winsorizing at 95%. These results can be found in the robustness checks section 4.2.A. This section also contains the normality results of the normal sample and the filtered sample.

Mean Median Standard Deviation T-statistic -5, +5 -1.60%* -0.90% 12.20% -1.9 -4, +4 -1.46%* -1.22% 11.96% -1.8 -3, +3 -1.53%** -1.36% 11.17% -2.0 -2, +2 -1.69%*** -0.97% 8.82% -2.8 -1, +1 -0.77%* -0.29% 7.94% -1.4 -0, +0 -1.15%*** -0.10% 6.22% -2.7 Banking Event Window

(22)

21

Since the mean returns and significance differ significantly by event window, the same steps are taken as in Section 4.1. First, the abnormal returns by day are graphed and then the prior- and post-event windows are shown.

Graph 4:

As can be seen, the abnormal returns that result from using the S&P 1500 or the NASDAQ Bank Composite differ greatly. In fact, the correlation between the S&P 1500 and NASDAQ Bank Composite during the (-5,+5) window is only 0.47. The correlation before the event day is 0.66 (window -5,-1) while the correlation after the event day is 0.06 (window +1,+5). The prior-event and post-event windows are shown on the next page:

(23)

22 Prior-event: Table 12: Notes: *,**,* ** repres ent signific ance at the 10%, 5% and 1% level. Table 13:

S&P 1500 Event Window

Mean Median Standard Deviation T-statistic -5, 0 0.58% 0.15% 10.64% 0.8 -4, 0 0.33% 0.03% 10.45% 0.5 -3, 0 0.50% 0.31% 9.77% 0.7 -2, 0 -0.65%* 0.30% 8.33% -1.1 -1, 0 -0.28% 0.42% 7.05% -0.6 Banking Event Window

(24)

23

Post-event:

Table 14:

As the results above indicate, the significance is high in banking event windows (-1,0), (0,+1), (0,+3), and (0,+4). Significance is the highest in the (0,+4) window. The S&P 1500 window shows a high significance in the post-event window that is higher than the significance in Table 12, indicating that the most of the negative abnormal returns are made in the post-event window. The banking post-event window also exhibits a high significance prior to the post-event date (windows -2,0 and -1,0). It might be that the banking industry had an information advantage and thus their returns adapted earlier to the event. It might also be that the market as a whole overestimated the values of the banks, and the capital injections functioned as a wake-up call to the market.

Mean Median Standard Deviation T-statistic 0, +5 -3.33%*** -1.85% 10.64% -6.4 0, +4 -2.94%*** -1.19% 10.45% -5.6 0, +3 -3.14%*** -0.91% 9.77% -6.0 0, +2 -2.20%*** -0.73% 8.33% -4.2 0, +1 -1.60%*** -0.15% 7.05% -3.1 Banking Event window

(25)

24 4.3. STATE EFFECTS

In the table below, the results of using the dummy states as an independent variable and the abnormal returns as a dependent variable in an OLS regressions are presented. The first table uses the abnormal returns that were calculated using the S&P 1500 as a market proxy. In this table, only the two variables that have the highest significance and the two variables that have the lowest significance are shown, this shown for multiple event windows. The type of event window can be found in the second row. There are approximately 43 states in the sample. S&P 1500, TARP Announcement Event Window

(26)

25 Tabl e 16:

Notes: *,**,*** represent significance at the 10%, 5% and 1% level.

As can be seen above, there are substantial differences in the values of the variables. For instance, the state of Idaho has a value of -0.395 (-39.5%) in the (-5, +5) window and is quite significant, while in the same window the variable Nevada has only a value of 0.016 and is not significant whatsoever. Additionally, the variables with the lowest values in the (-5,+5) window are not significant, indicating that the evidence for a state effect in these states is very low. Almost all the state variables are negative, meaning that a lower return is expected if one operates in certain states. Two states repeatedly appear in the event windows. These are Connecticut and Idaho, which in most cases have a significance level of 5%. In the smaller event windows (-1,+1) and (-0,+0) there is still a differences between the highest values for variables and the lowest values for variables, however the coefficients are not significant at a 10% level. Below are the results of using the abnormal returns from the capital injection as a dependent variable and the state dummy variables as an independent variable:

Table 17: MISSISSIPPI 0.003 12.94% 0.0 VIRGINIA -0.002 10.96% -0.0 S&P 1500, Capital Injection Event Window

(27)

26 Notes: *,**,*** represent significance at the 10%, 5% and 1% level.

Again, a large difference in the values of the coefficient for the different states can be seen, and this is present in all windows. However, significance at the 10% level or higher is not present in the (+1,-1) and (+0,-0) windows. As the state with the lowest coefficient, Minnesota repeatedly appears in most of the event windows, as does the state Kansas, which has the highest coefficient. The results using the banking abnormal returns during the general event window as the dependent variable are shown below. Again, the two variables with the highest significance and lowest significance are shown.

Table 18: PUERTO RICO -0.246** 12.96% -1.9 MINNESOTA -0.056 10.25% -0.5 MASSACHUSETTS -0.095 9.90% -1.0 -1,+1 PUERTO RICO -0.168 11.67% -1.4 SOUTH CAROLINA -0.126 9.04% -1.4 MINNESOTA -0.014 9.23% -0.1 GEORGIA -0.030 8.91% -0.3 -0,+0 DELAWARE 0.088 7.57% 1.2 MINNESOTA 0.115 6.91% 1.7 COLLORADO -0.004 8.74% -0.0 CONNETICUT 0.004 7.14% 0.0 Banking, TARP Announcement Event Window

(28)

27 Notes: *,**,*** represent significance at the 10%, 5% and 1% level.

Again, most states are not significant in the small event windows. Most states are only significant at the 10% level, and again great differences among the states can be seen. Connecticut is significant at a 10% level in most event windows, and even at a 1% level in the (-2,+2) window and (-3,+3) window.

Table 19: IDAHO -0.368** 15.04% -2.4 SOUTH DAKOTA -0.038 15.04% -0.3 ARKANSAS -0.016 15.04% -0.1 -2,+2 HAWAII -0.316 21.57% -1.5 CONNETICUT -0.193 17.61% -1.1 KENTUCKY -0.000 16.47% -0.0 VIRGINIA 0.001 15.83% 0.0 -1,+1 MAINE -0.192 16.07% -1.2 IOWA -0.169 15.15% -1.1 SOUTH DAKOTA -0.003 18.56% -0.0 OHIO -0.003 13.76% -0.0 -0,+0 UTAH 0.196 14.56% 1.3 OHIO 0.107 10.79% 1.0 NORTH CAROLINA -0.001 10.65% -0.0 KANSAS 0.003 14.56% 0.0 Banking Capital injection Type of Event

(29)

28 Notes: *,**,*** represent significance at the 10%, 5% and 1% level.

(30)

29 4.1.A. ROBUSTNESS CHECKS : TARP

This section shows the robustness checks. First shown are the filtered results. Due to filtering for outliers the means of the abnormal returns have become smaller and the T-values less significant. The results are presented below:

Table 20:

A Corrado test was also done, as this test does not suffer from the normality conditions. The results are shown on the next page and due to the parameters of the Corrado test are given only per day.

Mean Median Standard Deviation T-Statistic -5, +5 4.06%*** 3.00% 12.37% _-2.7 -4, +4 2.85%** 2.89% 11.67% _-2.0 -3, +3 7.48%** 8.08% 11.23% _-2.2 -2, +2 9.53%*** 7.24% 14.46% _-3.4 -1, +1 4.43%* 1.59% 10.40% _-1.7 -0, +0 4.30%*** 1.80% 8.21% _-3.5 Banking Event Window

(31)

30

Table 22:

Notes: *,**,*** represent

significance at the 10%, 5% and 1% level. Table 23:

Event Window S&P 1500

Mean T-Statistic Corrado

-5 0.29% 0.6 0.2 -4 -1.15% -2.5*** -0.7 -3 -1.39% -2.6*** -0.4 -2 4.82% 6.0*** 1.5* -1 -3.86% -4.4** -1.8* 0 4.61% 6.6*** 1.9** 1 4.14% 7.1*** 2.5*** 2 -0.08% -0.2 -0.8 3 -0.66% -1.6 -0.7 4 -3.49% -8.5*** -2.4*** 5 0.93% 2.7*** 1.2 Event Window Banking

Mean T-Statistic Corrado

(32)

31 Notes: *,**,*** represent significance at the 10%, 5% and 1% level.

As can be seen in Table 22 and Table 23, there is some overlap in

significance between the Corrado test and the T-test. Below are the normality results of Table 3, Table 4, Table 20 and Table 21. If the Jargue-Bera test is significant it indicates that normality is not present.

Table 24:

(33)

32 Notes: *,**,*** represent significance at the 10%, 5% and 1% level.

As can be seen after filtering for outliers, normality is improved in most of the windows, which was not present in the filter-free sample. The significance of the results in the winsorized sample is decreased, however significance is still present.

(34)

33 4.2.B. ROBUSTNESS CHECKS : CAPITAL INJECTIONS

Just as before, filtering for outliers results in the means of the abnormal returns becoming less negative and the T-values becoming smaller. The results are presented below:

Table 26:

Table 27:

Again a Corrado test was done, the results of which are shown on the following page.

Mean Median Standard Deviation T-Statistic -5, +5 -1.77%*** -0.88% _4.58% -2.4 -4, +4 -1.35%** -1.04% _6.68% -2.0 -3, +3 -1.36%** -0.96% _7.13% -2.2 -2, +2 -1.64%*** -0.98% _9.26% -3.0 -1, +1 -0.80%** -0.29% _9.91% -1.6 -0, +0 -1.10%*** -0.11% _9.66% -3.1 Banking Event Window

(35)

34

Table 28:

Table 29:

Banking Event Day

Mean T-STATISTIC CORRADO

-5 -0,12% -0.3 -0.3 -4 -0,08% -0.2 -0.2 -3 1,75% 3.9*** 3.0*** -2 -0,71% -2.1*** -1.6** -1 -0,51% -1.4* 0.0 -0 -0,98% -2.4*** -1.6** 1 0,35% 0.9 -0.5 2 -0,48% -1.5** -0.7 3 -0,39% -1.1 0.2 4 -0,33% -1.2 0.1 5 1,26% 2.3*** 0.2 S&P1500 Event Day

Mean T-STATISTIC CORRADO

(36)

35

Because significance is not present in all cases, a normality test of the filtered and normal sample was conducted. The results are shown below, if the Jargue-Bera test is significant normality is not present. It can be seen in the results in Table 29, the most significant days occur before the event date rather than after or on the event date. In contrast to this, in Section 4.2 the post-event tables (Table 14 and Table 15) showed more significance than the prior-event windows (Table 12 and Table 13). This discrepancy is likely because the mean on day 3 was positive, while on the other days before the event date (days -5, -4, -2, and -1) the mean was negative. The days after the event date are negative except for a small positive return on day 1. This probably the explanation that the Post-event windows are considerably more negative than the Prior event windows.

Table 30:

Table 31:

Again, it can be seen that normality is only present in the filtered samples. The significance of the results decline in comparison with the non-filtered sample.

(37)

36

5. CONCLUSION

This thesis asked the question, : “Were there any differences regarding the effectiveness of the Troubled Asset Relief Program across states in the United States of America based on stock returns? In order answer this, the first step was to measure the stock market response to the initiation of TARP and the capital injections under TARP using event study

methodology. Next, the abnormal returns from sub-sections 4.1 and 4.2 were used as dependent variables, and state dummy variables were used as independent variables in an OLS regression to measure the state effects.

Concerning the initiation of TARP and the capital injections under TARP, the results from sub-sections 4.1. and 4.2. are straightforward: a positive effect from the announcement of TARP and a negative effect from the capital injections under TARP were observed. As such the first two null hypothesis specified in the methodology section(page 12) are rejected in favour of the alternative hypotheses. This is to be expected if one considers the event study literature surrounding the TARP program. It is interesting to note that the choice of market proxy in the event studies makes a substantial difference in the significance of the returns and on which day the sign of the abnormal returns is positive or negative. To my knowledge, the papers that investigate events surrounding TARP use different market proxies from one another, varying from the S&P 500, to an industry benchmark or a self-constructed market proxy. Despite this, they do not test if their results change in any meaningful way if they use a different market proxy.

Concerning the state effects, from sub-section 4.3 a few results are remarkable. For

example, it is interesting that that the same states repeatedly appear, such as Kansas, Idaho, or Connecticut. Furthermore, when one considers the highest significance or lowest

significance variables in an event window, there are substantial differences in state effects. For example, in the (-5,+5) window of Table 16 (page 24), the coefficient of Idaho is almost 40 times as large as the coefficient of Nevada. As such the last 2 null hypotheses are reject in favour of the alternative hypotheses(page12).

(38)

37

6. LITERATURE REFERENCES

Bayazitova, D., and Shivdasani, A., 2012. Assessing TARP. Review of Financial Studies, 25(2), 377-407. Berger, A.N., Roman, R.A., 2015. Did TARP banks get competitive advantages?. Journal of Financial and Quantitative Analysis 50(06), 1199-1236.

Berger, A.N., Roman, R.A., and Sedunov, J., 2016. Do Bank Bailouts Reduce or Increase Systemic Risk? The Effects of TARP on Financial System Stability. Unpublished working paper. The Federal Reserve Bank of Kansas city.

Black, L.K., Hazelwood, L.N., 2013. The effect of TARP on bank risk-taking. Journal of Financial Stability 9(4), 790-803.

Blau, B.M., Brough, T.J., and Thomas, D.W., 2013. Corporate lobbying, political connections, and the bailout of banks. Journal of Banking & Finance 37(8), 3007-3017.

Duchin, R., Sosyura, D., 2014. Safer ratios, riskier portfolios: Banks׳ response to government aid. Journal of Financial Economics 113(1), 1-28.

Farruggio, C., Michalak, T.C. and Uhde, A., 2013. The light and dark side of TARP. Journal of Banking & Finance 37(7), 2586-2604.

Elyasiani, E., Mester, L.J. and Pagano, M.S., 2014. Large capital infusions, investor reactions, and the return and risk-performance of financial institutions over the business cycle. Journal of Financial Stability 11, 62-81. Fratianni, M.U., and Marchionne, F., 2010. The banking bailout of the subprime crisis: size and effects. PSL Quarterly Review 63(254) 185-231.

Fratianni, M., Marchionne, F., 2013. The banking bailout of the subprime crisis: Was the bang worth the buck?. Journal of International Financial Markets, Institutions and Money 23, 240-264.

Goldsmith-Pinkham, P., Yorulmazer, T., 2010. Liquidity, bank runs, and bailouts: spillover effects during the Northern Rock episode. Journal of Financial Services Research 37, 83-98.

Joines, A., 2010. Signals to the market: too big to fail banks and the recent crisis. Unpublished working Paper, University of Notre Dame.

King, M.R., 2009. Time to buy or just buying time? The market reaction to bank rescue packages. Unpublished working paper. Bank for International Settlements.

MacKinlay, A.C., 1997. Event studies in economics and finance. Journal of economic literature 35(1), 13-39. Ncube, M., 2016. Bank bailout under TARP in the US. Unpublished working paper. University of Oxford. Sieczka, P., Sornette, D., and Holyst, J.A., 2011. The Lehman Brothers effect and bankruptcy cascades. The European Physical Journal B 82, 257-269

Schoenmaker, D., 1996. Contagion risk in banking. LSE Financial Markets Group.

(39)

38 Veronesi, P., Zingales, L., 2010. Paulson's gift. Journal of Financial Economics 97(3), 339-368.

Weiß, G.N., 2012. Analysing contagion and bailout effects with copulae. Journal of Economics and Finance, 36(1), 1-32.

(40)

39

APPENDIX 1 LIST OF COMPANIES USED

Stock State Date

1 Bank of New York Mellon N.Y. Oct. 28, 2008

2 Citigroup N.Y. Oct. 28, 2008

3 Goldman Sachs N.Y. Oct. 28, 2008

4 JPMorgan Chase N.Y. Oct. 28, 2008

5 Morgan Stanley N.Y. Oct. 28, 2008

6 State Street Mass. Oct. 28, 2008

7 Wells Fargo Calif. Oct. 28, 2008

8 1st Financial Services Corp N.C. Nov. 14, 2008 9 Bank of Commerce Holdings Calif. Nov. 14, 2008

10 BB&T N.C. Nov. 14, 2008

11 Broadway Financial Corporation Calif. Nov. 14, 2008 12 Capital One Financial Corp. Va. Nov. 14, 2008 13 Comerica Incorporated Texas Nov. 14, 2008 14 First Horizon National Tenn. Nov. 14, 2008 15 Huntington Bancshares Ohio Nov. 14, 2008

16 KeyCorp Ohio Nov. 14, 2008

17 Northern Trust Ill. Nov. 14, 2008

18 Regions Financial Corp. Ala. Nov. 14, 2008

19 SunTrust Ga. Nov. 14, 2008

20 TCF Financial Minn. Nov. 14, 2008

21 U.S. Bancorp Minn. Nov. 14, 2008

22 UCBH Holdings Calif. Nov. 14, 2008

23 Umpqua Ore. Nov. 14, 2008

24 Valley National N.J. Nov. 14, 2008 25 Washington Federal Inc. Wash. Nov. 14, 2008

26 Zions Bancorp Utah Nov. 14, 2008

27 Ameris Bancorp Ga. Nov. 21, 2008

28 Associated Banc-Corp Wis. Nov. 21, 2008

29 Banner Corp Wash. Nov. 21, 2008

30 Boston Private Financial Holdings Mass. Nov. 21, 2008 31 Cascade Financial Corp Wash. Nov. 21, 2008 32 CenterState Banks of Florida, Inc. Fla. Nov. 21, 2008 33 City National Calif. Nov. 21, 2008 34 Columbia Banking System Wash. Nov. 21, 2008 35 First Community Bancshares Va. Nov. 21, 2008 36 First Community Corp S.C. Nov. 21, 2008

37 First Niagara N.Y. Nov. 21, 2008

38 First PacTrust Bancorp, Inc. Calif. Nov. 21, 2008 39 Heritage Commerce Corp Calif. Nov. 21, 2008 40 Heritage Financial Corp Wash. Nov. 21, 2008 41 HF Financial Corp S.D. Nov. 21, 2008 42 Pacific Capital Bancorp Calif. Nov. 21, 2008

43 Porter Bancorp Ky. Nov. 21, 2008

(41)

40

45 Taylor Capital Ill. Nov. 21, 2008

46 Trustmark Corp Miss. Nov. 21, 2008 47 Webster Financial Conn. Nov. 21, 2008 48 Western Alliance Bancorporation Nev. Nov. 21, 2008

49 CVB Financial Calif. Dec. 5, 2008

50 Bank of Marin Bancorp Calif. Dec. 5, 2008 51 Bank of North Carolina N.C. Dec. 5, 2008 52 Blue Valley Ban Corp Kan. Dec. 5, 2008 53 Cathay General Bancorp Calif. Dec. 5, 2008 54 Central Bancorp Mass. Dec. 5, 2008 55 Central Federal Corp Ohio Dec. 5, 2008 56 Coastal Banking Company Fla. Dec. 5, 2008

57 Eagle Bancorp Md. Dec. 5, 2008

58 East West Bancorp, Inc. Calif. Dec. 5, 2008 59 Encore Bancshares Texas Dec. 5, 2008 60 First Defiance Financial Corp Ohio Dec. 5, 2008 61 First Financial Holdings S.C. Dec. 5, 2008 62 First Midwest Bancorp Ill. Dec. 5, 2008

63 FPB Bancorp Fla. Dec. 5, 2008

64 Great Southern Bancorp Mo. Dec. 5, 2008

65 IBERIABANK Corp La. Dec. 5, 2008

66 Manhattan Bancorp Calif. Dec. 5, 2008

67 MB Financial Ill. Dec. 5, 2008

68 Midwest Banc Holdings Ill. Dec. 5, 2008 69 Oak Valley Bancorp Calif. Dec. 5, 2008 70 Old Line Bancshares Md. Dec. 5, 2008 71 Popular, Inc.

Puerto

Rico Dec. 5, 2008 72 Sandy Spring Bancorp Md. Dec. 5, 2008 73 South Financial Group S.C. Dec. 5, 2008 74 Southern Community Financial N.C. Dec. 5, 2008 75 Southern Missouri Bancorp Mo. Dec. 5, 2008 76 Southwest Bancorp Okla. Dec. 5, 2008 77 Sterling Financial Corp Wash. Dec. 5, 2008 78 Superior Bancorp Ala. Dec. 5, 2008 79 TIB Financial Corp Fla. Dec. 5, 2008 80 United Community Banks Ga. Dec. 5, 2008

81 Unity Bancorp N.J. Dec. 5, 2008

82 WesBanco W.Va. Dec. 5, 2008

83 Bank of the Ozarks Ark. Dec. 12, 2008

84 Capital Bank N.C. Dec. 12, 2008

85 Citizens Republic Bancorp Mich. Dec. 12, 2008 86 Citizens South Banking Corp N.C. Dec. 12, 2008 87 First Litchfield Financial Corp Conn. Dec. 12, 2008

88 HopFed Bancorp Ky. Dec. 12, 2008

89 Independent Bank Corporation Mich. Dec. 12, 2008 90 Indiana Community Bancorp Ind. Dec. 12, 2008

(42)

41 92 National Penn Bancshares Pa. Dec. 12, 2008

93 NewBridge Bancorp N.C. Dec. 12, 2008 94 Northeast Bancorp Maine Dec. 12, 2008 95 Old National Bancorp Ind. Dec. 12, 2008 96 Pacific International Bancorp Wash. Dec. 12, 2008 97 Pinnacle Financial Tenn. Dec. 12, 2008

98 Signature Bank N.Y. Dec. 12, 2008

99 Sterling Bancshares Texas Dec. 12, 2008 100 Susquehanna Bancshares Pa. Dec. 12, 2008 101 SVB Financial Group Calif. Dec. 12, 2008

102 The Bancorp Del. Dec. 12, 2008

103 TowneBank Va. Dec. 12, 2008

104 Valley Financial Corp Va. Dec. 12, 2008 105 Virginia Commerce Bancorp Va. Dec. 12, 2008 106 Wilmington Trust Corporation Del. Dec. 12, 2008 107 Wilshire Bancorp Calif. Dec. 12, 2008 108 Alliance Financial Corp N.Y. Dec. 19, 2008 109 AmeriServ Financial Pa. Dec. 19, 2008 110 Bancorp Rhode Island R.I. Dec. 19, 2008 111 BancTrust Financial Group Ala. Dec. 19, 2008 112 Berkshire Hills Bancorp Mass. Dec. 19, 2008 113 Citizens First Corp Ky. Dec. 19, 2008 114 CoBiz Financial Collorado. Dec. 19, 2008 115 Community Bankers Trust Corp Va. Dec. 19, 2008 116 Community Financial Corp Va. Dec. 19, 2008 117 Community West Bancshares Calif. Dec. 19, 2008 118 Connecticut Bank and Trust Company Conn. Dec. 19, 2008 119 Elmira Savings Bank N.Y. Dec. 19, 2008 120 Enterprise Financial Services Corp Mo. Dec. 19, 2008 121 Exchange Bank Calif. Dec. 19, 2008

122 FCB Bancorp Ky. Dec. 19, 2008

123 FFW Corp Ind. Dec. 19, 2008

124 Fidelity Southern Corp Ga. Dec. 19, 2008 125 First California Financial Group Calif. Dec. 19, 2008 126 Union First Market Bankshares Corporation Va. Dec. 19, 2008 127 Flushing Financial Corp N.Y. Dec. 19, 2008 128 Hawthorn Bancshares Mo. Dec. 19, 2008 129 Heartland Financial USA Iowa Dec. 19, 2008 130 Horizon Bancorp Ind. Dec. 19, 2008 131 Intermountain Community Bancorp Idaho Dec. 19, 2008 132 Mid Penn Bancorp Pa. Dec. 19, 2008 133 Monadnock Bancorp N.H. Dec. 19, 2008 134 Monarch Financial Holdings Va. Dec. 19, 2008

135 NCAL Bancorp Calif. Dec. 19, 2008

136 Patriot Bancshares Texas Dec. 19, 2008 137 Santa Lucia Bancorp Calif. Dec. 19, 2008 138 Seacoast Banking Corp Fla. Dec. 19, 2008

(43)

42 140 Summit State Bank Calif. Dec. 19, 2008

141 Synovus Financial Corp. Ga. Dec. 19, 2008 142 Tennessee Commerce Bancorp Tenn. Dec. 19, 2008 143 Tidelands Bancshares S.C. Dec. 19, 2008 144 Tri-County Financial Corp Md. Dec. 19, 2008 145 VIST Financial Corp Pa. Dec. 19, 2008 146 Wainwright Bank & Trust Mass. Dec. 19, 2008 147 Whitney Holding Corp La. Dec. 19, 2008 148 Wintrust Financial Corp Ill. Dec. 19, 2008 149 1st Constitution Bancorp N.J. Dec. 23, 2008

150 BCSB Bancorp Md. Dec. 23, 2008

151 Bridge Capital Holdings Calif. Dec. 23, 2008 152 Capital Pacific Bancorp Ore. Dec. 23, 2008

153 Cecil Bancorp Md. Dec. 23, 2008

154 Central Jersey Bancorp N.J. Dec. 23, 2008 155 Citizens Bancorp Calif. Dec. 23, 2008 156 Citizens Community Bank Va. Dec. 23, 2008 157 Community Investors Bancorp Ohio Dec. 23, 2008 158 Emclaire Financial Corp Pa. Dec. 23, 2008 159 Financial Institutions N.Y. Dec. 23, 2008 160 First Community Bank Corp of America Fla. Dec. 23, 2008 161 First Financial Bancorp Ohio Dec. 23, 2008 162 First Sound Bank Wash. Dec. 23, 2008 163 Fulton Financial Corp Pa. Dec. 23, 2008 164 Green Bankshares Tenn. Dec. 23, 2008

165 HMN Financial Minn. Dec. 23, 2008

166 International Bancshares Corporation Texas Dec. 23, 2008 167 Intervest Bancshares N.Y. Dec. 23, 2008 168 M&T Bank Corporation N.Y. Dec. 23, 2008 169 MutualFirst Financial Ind. Dec. 23, 2008 170 Pacific Commerce Bank Calif. Dec. 23, 2008 171 Park National Corporation Ohio Dec. 23, 2008 172 Parkvale Financial Corp Pa. Dec. 23, 2008 173 Peoples Bancorp of North Carolina N.C. Dec. 23, 2008 174 Seacoast Commerce Bank Calif. Dec. 23, 2008 175 Sterling Bancorp N.Y. Dec. 23, 2008 176 The Little Bank N.C. Dec. 23, 2008 177 Timberland Bancorp Wash. Dec. 23, 2008 178 Uwharrie Capital Corp N.C. Dec. 23, 2008 179 Fifth Third Bancorp Ohio Dec. 31, 2008 180 PNC Financial Services Pa. Dec. 31, 2008 181 West Bancorporation Iowa Dec. 31, 2008 182 American Express N.Y. Jan. 9, 2009 183 C&F Financial Corp Va. Jan. 9, 2009 184 Cadence Financial Corp Miss. Jan. 9, 2009 185 Carolina Bank Holdings N.C. Jan. 9, 2009

186 Center Bancorp N.J. Jan. 9, 2009

(44)

43 188 Centrue Financial Mo. Jan. 9, 2009

189 Codorus Valley Bancorp Pa. Jan. 9, 2009 190 Colloradony Bankcorp Ga. Jan. 9, 2009 191 Vantagesouth Bancshares, Inc. N.C. Jan. 9, 2009 192 Eastern Virginia Bankshares Va. Jan. 9, 2009 193 F.N.B. Corporation Pa. Jan. 9, 2009 194 Farmers Capital Bank Corp Ky. Jan. 9, 2009

195 First Bancorp N.C. Jan. 9, 2009

196 First Financial Service Corp Ky. Jan. 9, 2009 197 First Security Group Tenn. Jan. 9, 2009 198 FirstMerit Corp Ohio Jan. 9, 2009 199 Grandsouth Bancorporation S.C. Jan. 9, 2009 200 Independent Bank Corp Mass. Jan. 9, 2009 201 MidSouth Bancorp La. Jan. 9, 2009 202 Mission Community Bancorp Calif. Jan. 9, 2009 203 New York Private Bank & Trust Corp N.Y. Jan. 9, 2009 204 North Central Bancshares Iowa Jan. 9, 2009 205 Peapack-Gladstone Financial N.J. Jan. 9, 2009 206 Redwood Financial Minn. Jan. 9, 2009 207 Rising Sun Bancorp Md. Jan. 9, 2009 208 Security Business Bancorp Calif. Jan. 9, 2009 209 Security California Bancorp Calif. Jan. 9, 2009 210 Shore Bancshares Md. Jan. 9, 2009

211 Sun Bancorp N.J. Jan. 9, 2009

212 Surrey Bancorp N.C. Jan. 9, 2009