The Impact of TARP's Capital Purchase Program on the Stock Market Valuation of Participating Banks*

Transcription

1 The Impact of TARP's Capital Purchase Program on the Stock Market Valuation of Participating Banks* Jeffrey Ng Massachusetts Institute of Technology, Sloan School of Business 100 Main Street E Cambridge, MA, USA Florin P. Vasvari London Business School Regent s Park, London, NW1 4SA, United Kingdom fvasvari@london.edu Regina Wittenberg-Moerman University of Chicago Booth School of Business 5807 South Woodlawn Avenue, Chicago, IL, USA rwitten1@chicagobooth.edu This Draft: December, 2011 Abstract We examine the impact of the Capital Purchase Program (CPP), under the Troubled Assets Relief Program (TARP) initiated by the U.S. Treasury, on the equity market valuation of participating bank holding companies (CPP banks). Despite the Treasury s emphasis that the CPP s capital infusions were not a bailout of poorly performing banks, CPP banks were subject to low investor sentiment when the program was initiated. We find that CPP banks experienced significantly lower stock returns relative to non-participating bank holding companies (non-cpp banks) during the CPP initiation period, and that the equity market adjusted the valuation of the CPP banks upward in the quarters following the program s initiation. In contrast to their relatively lower equity market valuation when the program was initiated, we show that CPP banks had stronger fundamentals compared to non-cpp banks both prior to and during the program s initiation period. We also find evidence that negative media coverage exerted substantial downward pressure on banks stock returns and was associated with the significantly lower returns for the CPP banks relative to the non-cpp banks. Our results suggest that the investor sentiment significantly affected the CPP banks stock market valuation, contributing to the initial stock market undervaluation and subsequent correction. JEL classifications: E61, E63, G12, G14, G18, G21, M48 Key Words: Capital Purchase Program, Capital Infusion, Bank Performance, Valuation, Investor Sentiment, Media. * This paper was previously entitled The Participants in the TARP Capital Purchase Program: Failing or Healthy Banks? We wish to thank Ray Ball, Phil Berger, Michelle Hanlon, Steve Kaplan, Anil Kashyap, Christian Leuz, Scott Richardson, Sugata Roychowdhury, Doug Skinner, Ewa Sletten, Ross Watts, Rodrigo Verdi, participants at the 47th Annual Conference on Bank Structure and Competition at the Chicago Fed, and seminar participants at MIT and the University of Chicago Booth School of Business for their comments. We gratefully acknowledge the financial support of the MIT Sloan School of Management, London Business School RAMD Fund and the University of Chicago Booth School of Business.

2 1. Introduction The Troubled Assets Relief Program (TARP) was established under the Emergency Economic Stabilization Act approved in October 2008 with the specific goal of stabilizing the U.S. financial system and preventing a systemic collapse of the economy. TARP consisted of several programs, the most prominent of which was the Capital Purchase Program (CPP), announced on October 14, CPP was the program in which the U.S. Government, through the Department of the Treasury, infused capital into qualifying financial institutions with the objective of stabilizing the financial system. 2 Despite the Treasury s emphasis that the infusions were not a bailout of poorly performing banks, the program was the subject of much controversy, as it was not clear whether the program s participants were actually viable. In this paper, we examine whether the controversy regarding the financial health of banks participating in the CPP caused their valuation to diverge from fundamentals. Our analysis is motivated by the theoretical insights of DeLong, Shleifer, Summers and Waldmann (1990), who demonstrate that low investor sentiment leads to the undervaluation of stocks and divergence from fundamental values. Tetlock (2007) supports this prediction empirically and shows that high media pessimism exerts downward pressure on market prices followed by a reversion to fundamentals. Bank holding companies participating in the CPP (CPP banks hereafter) were subject to low investor sentiment during the program s initiation period due to the pessimistic public view of the CPP and its participant bank. 3 For example, during the 1 The U.S. government provided an extensive rescue package beyond the Capital Purchase Program; this included debt guarantees, short-term funding through Federal Reserve Bank facilities, the purchase of impaired assets and insurance against potential losses on specified portfolios of assets For instance, Stephen Wilson, chairman of LCNB National Bank in Ohio, argued that the the public perceives [participation in the CPP] as weakness and this is so discouraging because nothing could be further from the truth." (Satow, 2009). Douglas Elliott of the Brookings Institute reasoned that the TARP is "one of the most effective large-scale government programs that the public has vehemently decided was a bad idea" (Smith, 2010). 1

3 period when the program was initiated, the majority of articles in the Wall Street Journal about the CPP or its participant banks had a negative (pessimistic) tone. Supporting the importance of investor sentiment in the equity market valuation of banks, the model of Bernardo and Welch (2004) suggests that the pricing impact of investor sentiment is stronger when investors fear future liquidity shocks. During the period of the CPP s initiation, investors faced extremely high uncertainty with regard to their future liquidity needs. In addition, Baker and Wurgler (2006) show that the effect of investor sentiment on stock prices is more pronounced for firms that are more difficult to value. Relative to banks that did not take part in the program (non-cpp banks hereafter), participation in the CPP has significantly increased the uncertainty regarding the value of the CPP banks reported assets. Hoshi and Kashyap (2010) suggest that accepting the capital infusion could have signalled to the market that the receiving banks admitted to larger future losses than what they had previously disclosed. There was also high uncertainty with regard to the program s resolution mechanisms and the extent of the government s involvement in the affairs of the CPP banks. Further, the CPP infusions may have increased investors perception of the risk of future interventions and regulatory seizures (Bayazitova and Shivdasani, 2011). This heightened uncertainty significantly increased the difficulty of valuing CPP banks, further supporting their susceptibility to investor sentiment. To test whether participation in the CPP led to equity market undervaluation, we investigate the stock return performance of CPP banks relative to the stock return performance of benchmark non-cpp banks over the CPP initiation period (the last quarter of 2008 and the first quarter of 2009) and the post-cpp initiation period (the second, third and fourth calendar quarters of 2009). Controlling for the Fama-French (1992) risk factors, we find that the CPP bank portfolio significantly underperformed the non-cpp bank portfolio by 5.6% during the CPP 2

4 initiation period. However, over the post-cpp initiation period the equity market adjusted the valuation of the participating banks upward. During this period, the risk-adjusted return on the CPP bank portfolio was 10.3% higher than was the return on the non-cpp bank portfolio. 4 We verify that our results are not driven by banks that were forced to participate in the program in October We also show that the stock performance results are robust to controlling for momentum and cannot be explained by shifts in stock liquidity. We then investigate whether the relative stock return pattern of the CPP banks reflects their financial performance compared to non-cpp banks. We find that banks with higher profitability, a lower ratio of non-performing to total loans and a lower book-to-market equity ratio, as measured in the quarter preceding the program s initiation (i.e., the third quarter of 2008), were more likely to participate in the program. We find that banks with lower capital and cash-todeposits ratios were also more likely to participate in the program. These findings suggest that CPP participants had stronger fundamentals but experienced greater liquidity needs relative to non-cpp banks. We also verify that the performance of the CPP banks during the CPP initiation period was not inferior to that of the non-cpp banks. We find that CPP banks were in fact more profitable and reported lower non-performing loans ratio relative to non-cpp banks. Taken together, our findings suggest that, relative to the non-participating banks, the CPP banks had a stronger financial performance both prior to the program s initiation and during the CPP initiation period. These stronger fundamentals, coupled with capital infusions intended to enhance their strength, appear to be inconsistent with these banks relatively poor equity performance during the CPP initiation period. Therefore, we conjecture that the significantly 4 We restrict our sample to publicly traded bank holding companies with assets above $500 million. The CPP initiation period is the period during which recipient banks announced capital infusions. Of the 189 institutions with publicly traded equity and a size above $500 million that received capital infusions, 186 announced the capital infusion during the last quarter of 2008 or the first quarter of

5 worse stock performance of CPP banks relative to non-cpp banks indicates that the low investor sentiment that prevailed during this period may have caused these banks market valuation to diverge from their fundamentals. We find that CPP banks continue to be fundamentally stronger than non-cpp banks in the post-cpp initiation period, exhibiting higher profitability and lower non-performing loans ratios. However, this relatively better financial performance in the post-cpp initiation period is not significantly different from what it was during the CPP initiation period. Therefore, the reversal in the equity market valuation of CPP banks during the post-cpp initiation period cannot be attributed to changes in their relative performance. Rather, the return reversal suggests a reversion of the equity market s assessment of the CPP banks fundamental values, as investor sentiment improved. The improvement in investor sentiment in the post-cpp infusion period can be attributed to several factors such as the ability of CPP banks to pass the stress tests at the end of March 2009, early repayments of capital infusions starting from March 2009, as well as the banks ability to continue to make regular payments on the preferred shares issued to the Treasury. To shed more light on the association between investor sentiment and the stock return pattern of the CPP banks, we examine the tone of media coverage of the CPP and its participants. Tetlock (2007) and Tetlock, Saar-Tsechansky, and Macskassy (2008) show that media tone is strongly associated with investor sentiment and that the media can induce, amplify or reflect investor sentiment. 5 Following these studies, we focus on the coverage in the Wall Street Journal, which has the largest circulation of any financial publication in the United States. We classify articles related to the CPP over the CPP initiation and post-cpp initiation periods into 5 Because we are interested in the impact of investor sentiment on bank valuation, we do not examine whether the media induces or reflects investor sentiment. In addition, as noted in Tetlock (2007), it is empirically challenging to convincingly disentangle between media effects. 4

6 one of three categories: negative, positive and neutral. Negative (positive) articles are defined as news items that carry negative (positive) statements or implications about the CPP or the participant banks. 6 We find that during the CPP initiation period, the majority of the Wall Street Journal s articles had a negative tone (55%), indicating negative investor sentiment toward the program and participant banks. The negative media coverage dropped significantly by the end of the CPP initiation period in March 2009 and remained substantially smaller in the post-cpp initiation period relative to the prior period, averaging at 29%. In the regression analysis, controlling for the market beta, bank size and the book-to-market ratio, we find that the extent of negative media coverage is associated with substantially lower stock returns for both CPP and non-cpp banks. A one standard deviation increase in our Media measure, estimated on a monthly basis as the ratio of the negative articles to the total number of articles, translates into monthly bank returns that are 3.7% lower. We further show that the negative media coverage had a significantly stronger effect on the CPP banks. A one standard deviation increase in Media results in CPP banks monthly stock returns being lower by 1.77% relative to the non-cpp banks returns. These findings suggest that the tone of the media coverage can explain the significant underperformance of CPP banks relative to non-cpp banks during the CPP initiation period, when this coverage was predominantly negative. When the proportion of negative articles substantially decreased in the post-cpp initiation period, the return on the CPP banks reversed, adjusting the market s valuation. 7 We view these results as supporting our proposition that investor sentiment significantly affected the stock market 6 At least two of the co-authors have read each article and agreed on the classification. In total, we read 754 articles over the period from October 2008 to December 2009 (we provide more details in Section 4.4). 7 While it is also possible that the tone of the media coverage simply reflects the participant banks previous stock returns, we believe that this explanation is unlikely. The majority of the news items in the Wall Street Journal discussed the real effects of the CPP, such as stabilizing the economy and the extent of the increase in lending, in support of the positive or negative statements, views or implications expressed in the article. For a more extended discussion see Section

7 valuation of CPP banks, contributing to the initial undervaluation during the CPP initiation period and the valuation s subsequent reversal during the post-cpp initiation period. In addition, we perform a number of further tests on banks financial performance. We examine the frequency of bank delistings, given that delistings are typically the result of poor economic performance (Shumway, 1997). We find that CPP banks were less likely to delist, and to delist because of negative performance, in particular. This analysis further supports our proposition that CPP banks were healthier relative to non-cpp banks, due to both their stronger financial performance and the effect of the capital infusion. We also investigate the repayment of capital funds received under the CPP. If CPP participants were viable banks that experienced a temporary liquidity shortage, we expect the repayments to generate significantly positive returns to taxpayers. Consistent with our expectation, we find that as of March 31 st, 2010, taxpayers returns on the initial capital infusion and on the average capital infusion (the average of the original capital infusion and its remaining balance) were 9.96% and 15.26%, respectively. We contribute to the literature along several dimensions. First, we add to the literature on the role of investor sentiment in firm valuation (e.g., De Long et al., 1990, Baker and Wurgler, 2006, and Tetlock, 2007). To the best of our knowledge, our paper is the first to show that the low investor sentiment triggered by government intervention can negatively affect the valuation of firms that received capital infusions. By providing evidence that the valuation of CPP banks initially failed to capture their relatively stronger financial performance, we also highlight the importance of investor sentiment in banks valuation when financial markets are under stress. Second, we contribute to the literature on the role of the media in capital markets. The majority of previous studies show the positive consequences of media coverage, such as enriching the information environment, alleviating informational frictions and reducing the cost 6

8 of information acquisition (e.g., Miller, 2006, Dyck et al., 2008, Bushee et al., 2009, Fang and Peress, 2009, Li et al., 2010, and Soltes, 2010). In contrast, Shiller (2000) and Bhattacharya et al. (2009) suggest that the media hyped Internet stocks and consequently contributed to their dramatic rise and fall, while Solomon (2011) finds that positive media coverage excessively increases investor expectations, leading to eventual disappointment when a firm s hard information is revealed. Our findings that media coverage contributed to a temporary divergence of CPP bank values relative to their non-cpp peers, conditional on their financial performance, adds to the evidence on the potentially detrimental effect of media coverage on price formation in capital markets. Third, we contribute to the literature on government bailouts (e.g. Bernardo and Welch, 2004, Taliaferro, 2009, Hoshi and Kashyap, 2010, Veronesi and Zingales, 2010, Bayazitova and Shivdasani, 2011, Duchin and Sosyura, 2011). Bayazitova and Shivdasani (2011) find that better performing banks opted out of participating in the CPP. In contrast, Duchin and Sosyura (2011) provide evidence consistent with the program supporting stronger and more financially viable banks. Also, Taliaferro (2009) shows that, compared to non-cpp banks, CPP banks were more exposed to funding risk and had weaker capital ratios, but invested in healthier loan portfolios. Our findings that CPP banks had stronger fundamentals both prior to and following the CPP s initiation add to these studies and help to shed light on the primary factors that contributed to the CPP s successful implementation. The following section presents a brief description of the CPP. Section 3 describes the sample selection. Section 4 discusses our empirical findings. Section 5 concludes the paper. 2. Background on the Capital Purchase Program The Troubled Assets Relief Program (TARP) was established under the Emergency 7

9 Economic Stabilization Act, which was approved by the United States Congress on October 3, TARP, as originally envisioned in the fall of 2008, would have involved the purchase, management and sale of up to $700 billion of toxic assets, primarily troubled mortgages and mortgage-backed securities. This framework was quickly abandoned and the program s scope was changed to 12 announced programs that included capital infusions into banks, financing for the automotive industry, asset guarantees, the provision of affordable home loans, public-private investment programs, etc. The first and most prominent initiative under TARP, the Capital Purchase Program (CPP), was announced on October 14, Initially, the CPP was allocated up to $250 billion but, by the program s close on December 9, 2009, the Treasury had infused only about $205 billion in 742 transactions involving 709 financial institutions. 8 While the CPP aimed to increase the flow of financing to U.S. businesses and consumers and to support the U.S. economy, the U.S. government intervention via capital infusions into banks was quite controversial. On the one hand, the Treasury emphasized that the CPP was a voluntary program through which the government invested in healthy, viable institutions that were recommended by their applicable federal banking regulator to strengthen the capital base of these institutions and improve the stability of the financial system. 9,10 Consistent with this stated objective, for instance, Valley National Bancorp stated in a press release in October 2009 that it was pleased to announce that it has been chosen as one of the nation's stronger regional 8 The Treasury announced the establishment of another program, the Capital Assistance Program (CAP) in February 2009; its intent was to ensure that banks had a sufficient capital cushion to withstand larger than expected losses in the future. CAP included a stress test to evaluate capital buffers. If capital was needed and could not be raised from private markets, the banks would have been forced to accept CAP assistance in return for mandatory convertible preferred stock. On Nov. 9, 2009, the Treasury closed this program without making any investments (SIGTARP January 2010 Report). 9 See 10 Another program under TARP, the Target Investment Program (TIP), set up after the CPP, appeared to be more of a bailout of unhealthy banks. TIP was created to make investments in institutions that are critical to the financial system. When Bank of America and Citigroup needed additional capital infusions, more funds ($20 billion for Citigroup on December 31, 2008 and $20 billion for Bank of America on January 16, 2009) were provided under TIP. TIP was effectively closed on December 31, 2009, as both Citigroup and Bank of America had repaid the funds received. 8

10 banks to participate in the U.S. Treasury Department's TARP Capital Purchase Program. On the other hand, banking regulators and the Treasury did not publicly reveal the criteria used to approve the CPP applications, which amplified the uncertainty about the health of the financial institutions participating in the program. More importantly, the public and the media often characterized the capital infusion under the CPP as a government bailout of relatively weaker banks and a waste of taxpayers money. For instance, McCall Wilson, president of the Bank of Fayette County in Tennessee, said I knew the community at first would be upset because they perceived it [the infusion] as a bailout (Satow, 2009). Neil Barofsky, the inspector general for TARP, after reviewing the bailouts of ten big banks in October 2008, criticized Treasury officials for misleading the public over the health of some of these banks and for undermining popular trust in rescue efforts when lending did not increase (SIGTARP October 2009 Report). Consistent with these statements, we find that during the period when the program was initiated, the vast majority of articles in the Wall Street Journal about the CPP or its participant banks had a negative (pessimistic) tone (see Section 4.4 for a detailed description of the media analysis). 11 Under the CPP, capital was infused into qualifying financial institutions and investments were allocated so as to vary from one to five percent of the recipient s risk-weighted assets. 12 The investments involved the purchase of non-voting senior preferred shares; the Treasury demanded that these shares have a low initial dividend rate of 5% for 5 years and 9% thereafter. The purchase of shares also included 10-year warrants that provided the U.S. government with the option to purchase common stock for an amount equal to 15% of the preferred equity 11 Warren Buffett, in his 1990 Chairman s Letter to Berkshire Hathaway s shareholders, stated that The most common cause of low prices is pessimism - sometimes pervasive, sometimes specific to a company or industry. 12 Qualifying financial institutions include bank holding companies, savings associations and certain savings and loan holding companies. In this paper, we focus on bank holding companies. 9

11 infusion at a specified price in the future. The CPP received some modifications when the American Recovery and Reinvestment Act of 2009 was enacted in February These changes imposed more stringent executive compensation requirements and provided more flexibility for capital repayments by removing time restrictions and no longer requiring the banks to demonstrate that they had received private equity investment in proportion to the funds repaid. Although the eight largest investments of the program accounted for $134.2 billion, the CPP also had many more modest investments: 331 of the 709 recipients had received less than $10 million each by December 2009 (SIGTARP January 2010 Report). Except for eight financial institutions in our sample that were forced to participate in the CPP on October 14, 2008, all other institutions took part in the CPP voluntarily. 13 Selection for the CPP was driven partly by banks voluntary decision to submit an application but also by the Treasury s and direct banking regulators approval to participate in the program. These federal banking regulators, the Federal Deposit Insurance Corporation (FDIC) and the Federal Reserve Bank (FRB), evaluated all submitted CPP applications based on factors such as capital adequacy, liquidity, earnings and sensitivity to market risks; qualifying applications were sent to the Treasury for final approval. 14 A large number of banks withdrew their applications, but because the Treasury viability criteria were not made publicly available, it is not clear how many of them withdrew voluntarily despite being qualified or how many withdrew because they were advised by the banking regulator that they did not meet the requirements. Based on the number of applications processed and forwarded to the Treasury by July 2009, Taliaferro (2009) estimates 13 The capital infusion was initially provided to ten banks: Bank of America, Bank of New York Mellon, Citigroup, Goldman Sachs, JP Morgan, Merrill Lynch, Morgan Stanley, State Street, Wachovia Corporation and Wells Fargo. However, Merrill Lynch and Wachovia Corporation were acquired by Bank of America and Wells Fargo, respectively, and their capital infusions went to the acquiring institutions. The acquisition of both banks was completed by December 31, The application period for publicly held financial institutions closed on November 14, 2008; the application period for privately held institutions closed on December 8,

12 the FDIC s rejection rate to be around 11%, while the FRB s ranged from 20% to 39%. The Treasury approved most of the qualifying applications received from banking regulators. The repayment of the capital provided by the Treasury was subject to consultation with the appropriate federal banking agency. If the agency confirmed that a bank would have sufficient capital after repayment, the bank could pay back the entire CPP investment either in a lump sum or over time, as long as each payment was at least 25% of the original total investment (unless the last payment was less by default). When returning the CPP investment, banks also had the opportunity to repurchase the warrants received by the Treasury at their fair market value Sample selection We obtain data on CPP participation from the U.S. Treasury financial stability reports, which can be found at By December 31, 2009, 709 financial institutions had received a capital infusion under the CPP (Table 1, Panel A). We restrict our analysis to bank holding companies for two reasons. First, bank holding companies were the main recipients of the capital infusions. Second, our research design requires that financial institutions have similar characteristics and common financial reporting requirements. Because we retrieve financial reporting data from Consolidated Financial Statements for Bank Holding Companies FR Y-C (Call Reports), we limit our sample to bank holding companies (banks hereafter) with total consolidated assets above $500 million. 16 In addition, we analyze publicly traded banks because our tests employ equity market data. We also limit our sample to banks that 15 As of December 31, 2009, only three CPP recipients had declared bankruptcy: UCBH Holdings (November 24, 2009), Pacific Coast National Bancorp (December 17, 2009) and CIT Group (November 1, 2009). The Treasury does not expect to recover the infusions in these banks, which add up to about $2.6 billion (the largest loss, $2.3 billion, comes from CIT Group). 16 Bank holding companies with total consolidated assets of less than $500 million file a Call Report called Parent Company Only Financial Statements for Small Bank Holding Companies FR Y-9SP. The FR Y-9SP provides more limited data than FR Y-C, which prevents us from incorporating these companies into the analysis. 11

13 announced participation in the CPP during the last quarter of 2008 and the first quarter of 2009, excluding from the sample the three banks that announced their participation after this period. Our final sample includes 186 banks. Note that while the CPP banks in our sample represent only 26.2% of the total number of institutions that received capital infusions, their capital infusion make up 90.1% of the funds provided to banks under the CPP. Panel B of Table 1 presents the selection process for our control sample of 161 banks that did not announce their participation in the CPP by the end of the first quarter of Our research sample thus comprises 347 banks (full sample hereafter). Table 2 reports the distribution of the capital infusions and repayments over time. We present the distribution first by the announcement date (i.e., the date when the bank announced its participation in the CPP) and then by the commitment date (i.e., the date when the funds were transferred to the bank). 17 The last columns present the distribution of repayments and the net outflow-inflow of funds to the Treasury. Banks started repaying the CPP funds from March 2009, with the largest repayments in June and December For our sample, about 63.7% of the capital infusion has been repaid by the end of Bank executives often referred to the public s negative perception of the program as the primary motive for early repayments. 4. Results 4.1 Stock market performance of CPP and non-cpp banks In this section, we investigate the stock return performance of CPP banks relative to the non-cpp bank benchmark group during the CPP initiation and the post-cpp initiation periods (we obtain stock trading data from the CRSP database). We compute buy-and-hold stock returns 17 We collected the announcement dates from Factiva. For banks that did not have a press release about their capital infusion approval, we define the announcement date as the day of the relevant TARP transaction report. 12

14 on the portfolios of CPP and non-cpp banks based on the daily returns from the first day of each period to the last day of the period (equally weighted). If a bank delists in the interim, we include its delisting return. Table 3, Panel A presents descriptive statistics for the full sample and for each of the two bank portfolios. We also present the statistics separately for the CPP initiation and the post-cpp initiation periods. We start with a univariate analysis of the buy-and-hold returns on the CPP bank portfolio relative to the return on the non-cpp bank portfolio. This comparison is equivalent to an analysis of the industry-adjusted returns of the CPP banks, which is important to perform because during the credit crisis bank holding companies could have risk-return profiles that were significantly different from those of other firms. We find that the buy-and-hold returns of both bank groups are highly negative during the CPP initiation period (-41.7%), with CPP banks performing significantly worse relative to non-cpp banks. For this period, the return on the CPP bank portfolio was about 6% lower relative to the return on the non-cpp bank portfolio. However, during the post-cpp initiation period the equity market adjusted the valuation of the CPP banks upwards so that, by the end of December 2009, the buy-and-hold returns were significantly higher for CPP banks than for non-cpp banks by 14.3%. In addition, we observe a reversal in the returns for the CPP bank portfolio from -44.4% in the CPP infusion period to 10.8% in the post-cpp infusion period (the difference between these returns is significant at the 1% level and is reported in the last row of Table 3). While the return on the non-cpp bank portfolio also significantly improved, it remained negative in the post-cpp initiation period. In Panels B and C of Table 3, we estimate multivariate regressions that control for the Fama-French (1992) risk factors. Table 3, Panel B reports the comparative analysis of the returns in the CPP initiation period. Our primary specification is presented in Column (1). CPP is an 13

15 indicator variable that is equal to one if the bank is a CPP bank (a bank that announced its participation in the CPP by March 31, 2009), zero otherwise. The coefficient on CPP can be interpreted as the difference in the risk-adjusted returns between the CPP and non-cpp bank portfolios. We find that, controlling for market beta, bank size and book-to-market ratios, the CPP bank portfolio significantly underperformed the non-cpp bank portfolio by 5.6%. We verify that this result is not driven by the first eight banks that were forced to participate in the program in October As evidenced from Column (2), when we exclude these banks from the CPP bank portfolio, the returns of the remaining CPP banks continue to be significantly smaller than those of non-cpp banks. In Columns (3) and (4), we present robustness analyses where we substitute the CPP indicator variable by the CPP-amount variable. We define this variable as the amount of the capital infusion received by a bank as a percentage of the bank s total assets at the end of the third calendar quarter of CPP-amount takes the value of zero for non-cpp banks. Consistent with the findings in the first two columns, we find that the capital infusion is associated with significantly more negative returns for both the entire CPP bank portfolio and for the CPP bank portfolio that excludes the eight banks forced into the CPP. Economically, the coefficient on CPP-amount of in Column (3) indicates that a change of one standard deviation in the CPP-amount (equal to 1.263, see Table 3, Panel A) is associated with a more negative return of 3.54%. With respect to the control variables, we find negative and significant coefficients on Beta and Book-to-market. These factor loadings suggest that banks with higher market risk or that are more distressed performed worse during the CPP initiation period. Table 3, Panel C presents the analysis for the post-cpp infusion period. In sharp contrast to the results in the previous panel, we find that the CPP bank portfolio significantly outperforms 14

16 the non-cpp bank portfolio. In our main specification presented in Column (1), we find that the return on the CPP bank portfolio was 10.3% higher than it was on the non-cpp bank portfolio. This result is robust to excluding from the analysis the first eight banks that participated in the program and to using CPP-amount as an alternative variable of interest (see Columns (2) to (4)). In an untabulated analysis, we find that the coefficients on the CPP and CPP-amount in Panel C are significantly different from those in Panel B, emphasizing that the stock return reversal for the CPP bank portfolio is statistically significant during the post-cpp initiation period. The coefficients on the risk factors suggest that, in the post-cpp initiation period, larger banks experienced higher returns. 18 Taken together, the results presented in Table 3 indicate that CPP participation had a significant negative effect on banks equity valuation during the program s initiation period and that the equity market adjusted the valuation of CPP banks in the post-cpp initiation period. These results are consistent with the investor sentiment literature (e.g., DeLong et al., 1990, Baker and Wurgler, 2006, and Tetlock, 2007), which suggests that low investor sentiment, which likely prevailed during the CPP initiation period, led to an undervaluation by equity investors. During the post-cpp initiation period, as investor sentiment towards CPP banks potentially improved, the return on the CPP bank portfolio reversed, correcting the undervaluation. The improvement in investor sentiment can be attributed to CPP banks ability to pass the stress tests at the end of March 2009, the early repayments of capital infusions that started in March 2009, and the banks ability to make regular interest payments on the non-voting senior preferred shares issued to the Treasury. 18 The decrease in the number of observations for the post-cpp infusion period is due to banks that delisted prior to the start of this period. 15

17 We perform a number of additional tests that investigate the robustness of our results on the stock market performance. First, we control for momentum in both Panels B and C and find that the results remain the same. Second, we verify that the more negative performance of the CPP banks relative to the non-cpp banks during the CPP initiation period cannot be attributed to the House of Representatives Bill no passed on March 19, which imposed an additional 90% tax on bonuses received by employees of capital infusion recipients. In an untabulated analysis we find that CPP banks experienced positive returns in March 2009 (i.e., the negative cumulative return becomes smaller in March 2009). Third, we examine whether the relative CPP banks stock return pattern that we document in Table 3 can be explained by shifts in stock liquidity. For instance, if, during the CPP initiation period, CPP banks experienced a more significant drop in liquidity relative to the non-cpp banks, one would expect to see a substantially more negative stock return performance for the CPP bank portfolio. Similarly, if CPP banks experienced a stronger improvement in their stock liquidity relative to non-cpp banks during the post-cpp initiation period, liquidity changes could potentially explain the superior stock return performance of the CPP bank portfolio. Based on the bid-ask spread measure and the liquidity measure of Amihud (2002), and controlling for systematic risk factors, we find that during the CPP initiation period (the post-cpp initiation period), CPP banks did not experience a more significant drop (improvement) in liquidity relative to the non-cpp banks (untabulated). 19 This evidence indicates that stock liquidity changes across the two bank groups are unlikely to explain their relative equity return performance during and after the CPP initiation period. 19 We estimate the bid-ask spread measure as the average of the daily relative bid-ask spreads of a bank stock during the period. Amihud s (2002) liquidity measure is estimated as the average of the ratio of the daily absolute return to the daily dollar trading volume. 16

18 4.2. Banks fundamental performance In this section, we investigate whether the CPP banks stock return pattern relative to the non-cpp benchmark group reflects differences in the financial performance of these two groups of banks. We implement a series of tests that examine the relative differences in the performance measures between CPP and non-cpp banks prior to and following program s the launch Bank fundamentals associated with CPP participation In our first analysis, we investigate which bank characteristics are associated with participation in the CPP. We estimate the following logistic regression specification: CPP = β 0 + β 1 ROA-ytd + β 2 NPL + β 3 Capital-adequacy + β 4 Cash-to-deposits + β 5 Uninsured-deposits + β 6 Fair-value-exposure + β 7 Interest-sensitivity + β 8 Assets + β 9 Population + β 10 GDP growth + β 11 Unemployment + β 12 Blue-state + ε (1) CPP is an indicator variable that equals one if the bank announced its participation in the program during the CPP initiation period, zero otherwise. The main bank fundamental characteristics that we examine capture the bank s viability with respect to profitability and the quality of the loan portfolio. We measure profitability by the ratio of year-to-date net income to total assets (ROA-ytd) and the quality of the loan portfolio by the ratio of non-performing loans to total loans (NPL). Non-performing loans are loans that are past due 30 days, 90 days or noninterest-accruing. In an additional specification, we replace the ROA-ytd and NPL with an alternative measure of bank viability, Book-to-market. Book-to-market, which is the ratio of the book value of equity to the market value of equity, is a possible indicator of financial distress that incorporates market-based inputs. That is, banks whose market value of equity is lower 17

19 relative to their book value of equity are likely to be in greater financial distress. The market value of equity, measured at the end of September 2008, is obtained from the CRSP database. We also add controls for the long-term and short-term bank liquidity using Capitaladequacy and Cash-to-deposits, respectively. Capital-adequacy is the ratio of total capital to risk-weighted assets and Cash-to-deposits is the ratio of cash to total deposits. We measure the riskiness of the bank using Uninsured-deposits, Fair-value-exposure and Interest-sensitivity. Uninsured-deposits is the ratio of time deposits of $100,000 or more to total deposits, 20 Fairvalue-exposure is the ratio of available-for-sale securities and trading securities to total assets, and Interest-sensitivity is the ratio of the absolute value of net assets that are sensitive to shortterm interest movements to total assets. We also control for the size of the bank; Assets is the logarithm of total assets (total assets in billions). All financial statement characteristics are measured using call reports for the fiscal period ending in September We also incorporate into the analysis state-level characteristics, as economic and political factors within the state might influence a bank s participation in the CPP (e.g., Li, 2010, and Duchin and Sosyura, 2011). Population is the logarithm of the population (measured in millions) in the state where the bank is headquartered; this data is obtained from the Census Bureau. GDPgrowth is the percentage change in the gross domestic product of the state where the bank is headquartered in 2008; this data is obtained from the Bureau of Economic Analysis. Unemployment is the percentage of unemployment in 2008 in the state where the bank is headquartered; this data is obtained from the Bureau of Labor Statistics. Finally, Blue-state is an 20 We define this measure based on the $100,000 threshold for uninsured deposits, which was the Federal Deposit Insurance Corporation s (FDIC) threshold on September 30th, The FDIC increased the deposit threshold up to $250,000 in October Because call reports continue to provide uninsured deposits data based on the $100,000 threshold, we use this definition in all of our tests. 18

20 indicator variable that equals one if, in the 2008 presidential elections, the Democrats won in the state where the bank is headquartered. Table 5, Panel A reports the means and standard deviations of the fundamental characteristics for all sample banks, as well as a comparison of the means for CPP and non-cpp banks. In the quarter prior to the program s initiation, on average, sample banks had a return-onassets of zero, a ratio of non-performing loans to total loans of 3.4% and a book-to-market ratio of The average capital adequacy, as measured by the total risk-based capital ratio, is 12.6%; this figure is substantially above the 8% minimum capital requirement. Based on univariate tests of differences in means, CPP banks show a stronger performance across a few dimensions. CPP banks had a significantly higher profitability, a lower ratio of non-performing loans and a lower book-to-market ratio. CPP banks, however, had significantly weaker risk-based capital ratios. In addition, CPP banks were substantially bigger than the non-cpp banks. Table 5, Panel B presents the multivariate analysis of Equation (1). 21 In Column (1), the positive and statistically significant coefficient of on ROA-ytd suggests that more profitable banks are more likely to participate in the CPP. 22 An increase of one standard deviation in bank profitability increases the probability of participation in the program by 14.5%. We also find that banks with a higher ratio of non-performing loans (NPL) have a lower likelihood of participating in the CPP. An increase of one standard deviation in NPL decreases the probability of participation by 11.0%. In addition, banks with higher capital ratios (Capital- 21 There are 341 banks in this analysis. The decrease in the number of observations relative to the 347 banks used in the stock return analysis is due to the following exclusions. First, we exclude Goldman Sachs and Morgan Stanley because these banks did not file call reports for the period ending Q as they only became bank holding companies on September 21, Second, four banks are headquartered in Puerto Rico and therefore are missing state-level characteristics. 22 In unreported tests, we further analyze whether CPP-participating banks engaged in earnings management to boost their profitability and improve their perceived healthiness. We look at discretionary loan loss provisions recorded by these banks during the period prior to CPP initiation and their change during the CPP participation period and find no evidence of earnings management. These analyses are available upon request. 19

21 adequacy) and liquidity (Cash-to-deposits) are less likely to participate in the program; an increase of one standard deviation in Capital-adequacy and Cash-to-deposits decreases the probability of participation by 47.5% and 31.5%, respectively. Finally, we show that banks with higher uninsured deposits ratio (Uninsured-deposits) and larger banks (Assets) are more likely to participate in the CPP. When we substitute profitability and non-performing loan measures with Book-to-market in Column (2), we find that banks that are relatively more financially distressed are less likely to participate in the program. An increase of one standard deviation in Book-tomarket reduces the probability of participation by 48.7%. Overall, our findings suggest that CPP participants had a stronger financial performance, but experienced greater liquidity needs. This evidence supports the Treasury s claim that the capital infusions were made with the purpose of strengthening the capital base of healthy financial institutions. The stronger performance of CPP banks relative to non-cpp banks that we document is inconsistent with the common perception of CPP participants as unviable institutions. We perform a number of robustness tests. First, in the last two columns of Panel B, we demonstrate that our results are robust to the removal from the CPP sample of the first eight banks that were forced to participate in the CPP. Second, to address the frequently expressed criticism that the CPP was actually a bailout of institutions that were too big to fail, we exclude from the sample banks that had to comply with stress tests, as defined by the Treasury on February 10, This ex-post risk measure captures banks which, according to the Treasury, present systemic risk and, therefore could be subject to different decision criteria in the CPP approval process. In untabulated analyses, we find that our results are robust to the exclusion of 20

22 fifteen systematically important banks from our initial sample. 23 Third, we reclassify as CPP banks 22 banks that had received approval for, but rejected, the infusion. Although these banks did not receive a capital infusion, they might be very similar to the CPP banks and thus should not be part of the non-cpp sample. Similarly, we remove from the control sample 28 banks that publicly announced that they would not be applying for an infusion under the CPP. 24 Our findings are robust to these different specifications of the CPP and non-cpp bank samples (untabulated) Bank fundamentals following the CPP initiation To further examine whether the market valuation of the CPP banks compared to the non- CPP banks reflected their relative performance, we analyze bank performance following the program s initiation. We investigate differences in profitability and the non-performing loans ratio between the CPP and non-cpp banks by the following OLS regressions: ROA / NPL = β 0 + β 1 CPP + β 2 Capital-adequacy + β 3 Cash-to-deposits + β 4 Uninsured-deposits + β 5 Fair-value-exposure + β 6 Interest-sensitivity + β 7 Assets + β 8 Population + β 9 GDP growth + β 10 Unemployment + β 11 Blue-state + ε (2) ROA is measured by the ratio of net income to total assets; NPL is the ratio of nonperforming loans to total loans. Both measures are averaged across the CPP initiation period (e.g., NPL is the average measure of the non-performing loans ratios in the fourth quarter of 23 Stress tests under the Capital Assistance Program were imposed on nineteen institutions, but four of them are not included in our analysis. Goldman Sachs, Morgan Stanley, American Express Co. and GMAC LLC became bank holding companies in the fourth quarter of 2008 and therefore do not have Call Reports for the September 2008 quarter. 24 To identify these banks, we performed a Factiva search for all public announcements of 161 non-cpp banks over the period from October 2008 to March We have separately searched for articles on each of the banks in the control sample and used the following keywords to identify the relevant press releases or articles: Troubled Asset Relief Program or Bailout, TARP, Capital Purchase Program, CPP. 21

23 2008 and the first quarter of 2009). All the other variables are defined in Section We do not examine the book-to-market ratio as the dependent variable in this analysis because of the potential mispricing of banks equity following the program s initiation. Results presented in Table 6, Panel A reveal that, during the CPP initiation period, CPP banks experience a significantly lower non-performing loans ratio relative to non-cpp banks (see Column (1)). 25 The CPP banks ratio of non-performing loans to total loans is lower by Economically, this difference represents 35.4% of the mean non-performing loans ratio for the full bank sample. We note that this result is not due to an increase in the lending activity of the CPP banks relative to the non-cpp banks. In untabulated analysis, we find that the lending activity of CPP banks was not different from that of non-cpp banks during the CPP initiation period. This evidence is consistent with the findings of Taliaferro (2009), who documents that the bulk of the capital received by CPP banks has been used primarily for increasing regulatory capital ratios and that only a small fraction of it has been allocated to support increased lending. Similarly, CPP banks demonstrate higher profitability during the CPP initiation period (see Column (2)). The return-on-assets for CPP banks is higher than it is for non-cpp banks. This difference is economically significant as the mean profitability ratio for the full research sample is We find similar results when we eliminate those banks required to participate in the CPP (the last two columns of Table 6, Panel A). For comparison, in Table 6, Panel B we present the results for the estimation of banks fundamental performance in the post-cpp initiation period. ROA and NPL measures are now averaged over the post-cpp initiation period. We find that the CPP banks continue to be fundamentally stronger than the non-cpp banks: CPP banks have significantly lower non- 25 We use 333 banks in this analysis. Compared to the 341 banks used in the Table 5 analysis, eight non-cpp banks were excluded because of missing data from call reports for the fourth quarter of 2008 or the first quarter of