Payback Time for US Banks

(1)

Payback Time for

US Banks

A cross-sectional analysis of the relation between CEO

compensation and speed of government support repayment

(2)

Administrative Data

Rijksuniversiteit Groningen Faculty Economics & Business Msc. BA Finance

Name Michiel Dijkstra

Student number 1386085

Month and year August 2010

Thesis title Payback Time for US Banks: A cross-sectional analysis of the relation between CEO

compensation and speed of government support repayment

Adviser Prof. Dr. R.A.H. van der Meer

Second Supervisor Dr. A. Plantinga

Abstract In this research, the link between CEO compensation before and after the crisis and post-crisis performance is researched. Post-crisis performance is indicated by the time the banks have been in the Troubled Asset Relief Program (TARP), a government program to support the banking sector. In this program, the Treasury buys preferential shares in the bank, which have to be repurchased by the bank to exit the program. Constructing a random sample of 164 participating banks, I find a positive relation between performance-based CEO compensation before the crisis and the speed with which the TARP funds are repaid, consistent with agency theory. The sample consists only of TARP recipients, which are generally well-functioning banks. It can therefore be carefully stated that, if a company has an adequate corporate governance framework with respect to risk-taking,

performance-based CEO compensation increases

(3)

(4)

1. Introduction

“When the music stops, in terms of liquidity, things will be complicated. But as long as the music is playing, you’ve got to get up and dance. We’re still dancing.”

- Charles “Chuck” O. Prince III, CEO of Citigroup July 10, 2007

“Your company is now bankrupt, our economy is in a state of crisis, but you get to keep $480 million. I have a very basic question for you. Is this fair?”

“[…] I believe my cash compensation was close to $60 million […]. And I believe the amount that I took out of the company over and above that was, I believe, a little bit

less than $250 million. Still a large number, though.”

- Rep. Henry Waxman and Richard Fuld Jr., former CEO of Lehman Bros. October 7, 2008

"[The $700 billion rescue plan is] not based on any particular data point. We just wanted to choose a really large number."

- A Treasury spokeswoman September 19, 2008

As the causes and consequences of the credit crisis became increasingly clear, public resentment of banks – and in particular Wall Street banks – grew. The exchange between Rep. Waxman and Richard Fuld above perfectly illustrates this resentment, and highlights the outrage over the generous compensation packages that the CEOs of the troubled banks received. On top of that, most of these banks – those that did not go under – received government aid in the form of the Troubled Asset Relief Program (TARP).

The TARP was signed into law on October 3, 2008 after the bankruptcy of Lehman Brothers sent the markets into a downward spiral, causing liquidity to dry up and lending – especially interbank lending - to stop. The TARP was designed to combat this turmoil, as it was feared that the economy might go down with the fortunes of the financial sector.

(5)

the other banks, participation was voluntary, provided their financial health was adequate.

Starting out as an emergency measure to aid the ailing financial sector, TARP has become somewhat of a political instrument. Legislation passed after the distribution of TARP funds imposed extra conditions on TARP recipients. Most notably, this included an executive payment cap, ostensibly as a punitive measure. This is understandable in light of the furor over the high compensation levels of CEOs, but is it also judicious in an economic sense?

The high compensation came with performance-based pay, which is designed to lower agency costs. These arise due to the fact that the managers of a firm have different objectives than the owners (Jensen and Meckling, 1976). Performance-based pay aligns the objectives of the managers with those of the owners. The partly performance-based compensation packages have become very generous, also in recognition of the fact that CEOs possess unique and highly sought-after skills, which are valued accordingly (Hubbard and Palia, 1994).

In hindsight, it is tempting to suppose that the compensation packages have failed to properly align the objectives of manager and owner and indeed that the sheer magnitude of compensation in the banking sector has led to perverse incentives and may have caused the crisis1. However, banks are sensitive to systemic risk, which means that a crisis hits all banks, although those that are managed badly will generally be hit harder than others. Because banks are so vital to the economy, they almost always receive government support in times of crisis (Reinhart and Rogoff, 2008).

TARP was no exception, but it stands out in its financial and legislative scope. The government-imposed cap on executive compensation is especially notable, and its effects on the whole proceedings of TARP are quite interesting. It is thought that the speed with which some banks have returned the money to the US government can be attributed to the effects of the payment cap.

(6)

I research the relation between CEO compensation, both before and during the rescue program, and the time it took banks to repay the government support. The TARP program was host to a set of extraordinary conditions, and this research offers a new perspective on the relation between performance-based executive compensation and company performance in these circumstances. Already, research has come out concerning the effects of executive compensation on the crisis (Fahlenbrach and Stulz, 2009 and Cheng, Hong and Scheinkman, 2009) which connect the banks’ performance during the credit crisis with executive compensation. In this paper, I connect the banks’ CEO compensation to post-crisis performance, as evidenced by the time they take to repay the TARP funds.

Repaying the TARP funds as soon as possible is an economically sound decision (Bayazitova and Shivdasani, 2009). I find that, controlling for size and the financial condition before the crisis, a higher performance-based percentage of the CEO’s compensation before the crisis is associated with a lower time taken to repay TARP. This evidences that performance-based CEO pay is conducive to better company performance, even after a systemic crisis.

(7)

2. Problem Statement

The reason for researching the link between CEO compensation and post-crisis performance is twofold. First, in the wake of the crisis, the public perception has arisen that the CEOs of banks have received very large amounts of money for running their banks, and the economy, into the ground. Second, there is a large body of research into the compensation of executives, specifically the link between performance-based pay and company performance. Researching this relation in the context of a crisis, where the companies received government support, is interesting for both the uniqueness of the situation and the public perception of the CEO compensation.

Problem statement: does the level of CEO compensation before and after the crisis have a positive effect on post-crisis performance?

The post-crisis performance will be indicated by the time the banks take to pay back the government funds. The specifics of financial institutions, CEO compensation and the TARP program will be discussed. The problem statement will be answered using the following research questions.

1. What qualities distinguish financial institutions from other companies? 2. How do banking crises come about and how do banks function in crises? 3. How does the compensation of CEOs come about?

4. What are the characteristics of the Troubled Asset Relief Program (TARP)? 5. What effects does TARP have on individual banks?

6. What are the reasons for exiting the TARP program?

(8)

3. Theoretical Foundations

Here the theoretical foundations for the research will be provided. I start with an introduction of the business and characteristics of financial institutions, before discussing the particulars of these characteristics: liquidity risk and the role of a bank’s capital. Subsequently, I discuss CEO compensation, agency problems and the possible link with the financial crisis. This chapter will be ended with an explanation of the Troubled Asset Relief Program (TARP) and an overview of the literature on TARP so far.

3.1 On financial institutions

The place of financial institutions in an economy is to make the financial markets more efficient through its ability to provide market knowledge, transaction efficiency and contract enforcement. They facilitate the flow of money through the economy. To quote Oldfield and Santomero (1997): ‘If savers and investors, buyers and sellers, could locate each other efficiently, purchase any and all assets costlessly, and make their decisions with freely available perfect information, then financial institutions would have little scope for replacing or mediating direct transactions’. Financial institutions can broadly be divided into three groups:

1. Deposit-taking institutions that accept and manage deposits and make loans – includes banks, savings and loans associations, credit unions, trust companies and mortgage loan companies;

2. Institutions that manage assets to cover uncertain liabilities - includes insurance companies and pension funds;

3. Institutions that offer companies and individuals access to financial markets – includes brokers, underwriters and investment funds.

The first group, deposit-taking institutions (hereafter: banks), are worthy of closer inspection, as they possess the following three economic functions2_:

(9)

1. Credit intermediation

Rather than consumers and companies trying to lend and borrow to one another, banks act as middle men by lending and borrowing on their own account.

2. Credit quality improvement

Banks are much more creditworthy than the individual parties to which they have lent money to. This is because banks hold a diversified portfolio of loans.

3. Maturity transformation

Mostly, a bank’s liabilities are on a much shorter term than its assets: money on current accounts can be withdrawn at a moment’s notice, while a loan cannot be amortized immediately. Because of the size of its borrowed portfolio and its creditworthiness, a bank is able to bridge the maturity gap.

3.2 Liquidity risk

Banks’ ability to transform illiquid assets into more liquid liabilities presents a risk to the bank: the risk that it cannot satisfy the demand for liquidity from its deposit holders. Deposit holders can withdraw their deposits at any time or on very short notice. When a bank fails to supply deposit holders with liquidity, this will induce more deposit holders to liquidate their deposit, leading to a bank run (Diamond and Dybvig, 1983). A suspicion of insolvency will greatly increase the demand for liquidity amongst deposit holders, thus making the suspicion a self-fulfilling prophecy (Merton, 1968). Liquidity risk makes banks inherently unstable. In fact, Diamond and Rajan (2001) argue that this financial fragility is the key to liquidity creation.

(10)

Even fear of illiquidity may have adverse economic effects. Bebchuck and Goldstein (2009) show that if companies are dependent upon other companies’ financing ability, a credit freeze may arise when banks refuse to lend on the (self-fulfilling) expectation that other banks refuse to lend as well. Bernardo and Welch (2004) propose a model in which risk-neutral investors fear having to liquidate after a liquidity run, before prices return to fundamental values. Instead of doing so, the investor may prefer to sell today, thus triggering a liquidity run. In this model, liquidity crises are not caused by liquidity shocks per se, but by the fear of future liquidity shocks.

3.3 Banks’ capital adequacy and bank failure

There are extensive guidelines for the level of capital a bank requires (Berger, Herring and Szego, 1995). Why is this so important? As banks make money on the difference between borrowers and deposit holders, they have a high leverage. A bank can easily increase its income by increasing its leverage. This does make it more sensitive to economic shocks, so there is a tradeoff to be made here. Merton (1993) describes a bank’s equity capital as a ‘Wonderful, all-purpose cushion for absorbing risks of the institution, [because] management need not know what the source of the unanticipated loss is. […] Equity protects the firm against all forms of risk’. He points out that equity capital is quite expensive exactly for that reason, because of agency costs and taxation.

Santomero and Watson (1977) model the probability of default as a function of capital ratio, and find that the socially optimal capital ratio is a tradeoff between the probability of bank failure and inefficiency arising out of overcapitalization. Empirically, this is supported by Goddard, Molyneux and Wilson (2004b), who conclude that banks with a higher capital ratio grow slower and record lower profitability. They also find that current profit is an important prerequisite for future growth. This is due to the occurrence of persistence of profit in the banking sector (Berger, Bonime, Covitz and Hancock, 2000).

(11)

Bank failures are caused by a combination of unfavorable macroeconomic conditions and the bank’s weakness to withstand such conditions (Gavin and Hausmann, 1996). This weakness is attributed to poor management by Barr, Seiford and Siems (1994). González-Hermosillo (1999) investigates individual bank failures. Controlling for macroeconomic and regional factors, she finds that probability of failure is positively related to: significant property-related lending and reliance on interbank funding and negatively related to the amount of tradable securities on the balance sheet and loan yield. She also finds that contagion plays a significant role in bank failures, in line with Diamond and Rajan (2005).

3.3 Agency problems and CEO compensation

The literature on CEO compensation is largely founded upon the seminal paper by Jensen and Meckling (1976), “Theory of the Firm”. The authors show that the executive of a firm (the agent) has different objectives than the owner of the firm (the principal), so he would not lead the firm so well as if it were his own. This leads to agency costs, which can be lessened by aligning the interests of the agent and the principal. This involves making the payment of the agent dependent upon the state of wealth of the principal, which can be done by giving executives performance-based bonuses, stock options or stock in the firm.

CEOs compensation packages generally consist of four elements: base salary, an annual bonus, stock options and long-term incentive plans (including restricted stock plans) (Murphy, 1999). The base salary is based on benchmarking. The bonus is related to an accounting performance figure, such as EBITDA, and usually varies between a minimum (if the performance figure is more than a minimum) and a maximum (if the performance figure is higher than the target figure). Stock options typically become “vested” (exercisable) over time.

(12)

large enough for the CEO to invest the firm’s resources in valuable – but risk-increasing projects that he otherwise would have passed on.

This is especially relevant for the banking sector. Bebchuck and Spamann (2009) show that the particular combination of compensation structure and capital structure may induce bank executives to take excessive risks. Aso, they are paid better than CEOs in other sectors of the economy (Hubbard and Palia, 1994). Hubbard and Palia report a positive relation between executive pay and bank performance.

The notion that performance-based CEO compensation begets high firm performance is not without controversy. It is estimated by Tosi, Werner, Katz and Gomez-Mejia (2000) that only five percent of the variance in CEO pay can be attributed to firm performance. There is also evidence that CEOs are paid for just being lucky, especially in poorly governed firms (Bertrand and Mullainathan, 2001).

So, the evidence of the effectiveness of performance-based pay is mixed to say the least. As an explanation, Bebchuck, Fried and Walker (2002) suggest that managers have too much power in the decision concerning their compensation, which leads to less than optimal compensation structures from the owners’ point of view.

Blinder notes that the large compensation the bank CEOs received created perverse incentives and may have caused the crisis3. Fahlenbrach and Stulz (2009) find no evidence that banks performed better during the crisis when the incentives of CEOs were better aligned with the interests of shareholders. A positive correlation between performance-based compensation and risk-taking is found by Cheng, Hong and Scheinkman (2009). Firms with high executive payment showed extreme good performance pre-crisis and extreme bad performance during the crisis.

Within a broader context, the supposed failure to align manager’s and owner’s objectives via the compensation structure is also seen as a failure of corporate governance (Kirkpatrick, 2009). The combination of the presence of high performance-based compensation and the absence of proper risk management,

3_{(Blinder, Alan S.), “Crazy Compensation and the Crisis”, Wall Street Journal, May 28, 2009, link:}

(13)

Kirkpatrick argues, was a major contributor to the crisis. He notes that both Lehman Brothers and Bear Stearns had very poor risk oversight4.

3.4 Government intervention

Crises are endemic to the banking sector because of the particular characteristics of banks. Many banking crises have occurred over the history of banking (Caprio and Honohan, 2008). In that sense, the last one was not unique. Banks often receive government support in times of crisis (Reinhart and Rogoff, 2008).

Hoggarth, Reis and Saporta (2002) estimate the economic costs of systemic banking crises to be about fifteen to 20 percent of the GDP. So, government has a good reason to intervene in the banking sector. In fact, it is uniquely qualified to do so. Gorton and Huang (2004) argue that when the banking system becomes insolvent, the amount of liquidity needed to make the banks solvent again is so large that it cannot be provided by private agents – holding these liquid assets beforehand would have been economically inefficient. The government can improve welfare by creating this liquidity. This does open the door to moral hazard, as banks can take on too much risk as they know the government will bail them out (Goodhart and Huang, 2000).

Diamond and Rajan (2002) argue, based on the mutual influence between individual bank solvency and aggregate liquidity, that a government should bailout strong banks, as these are capable of supplying more liquidity to the system. The bailing out of weak banks would not cure the problem and could lead to an escalating sequence of bailouts, until the government has no more resources.

3.5 TARP program

To stabilize the financial system, Treasury Secretary Henry Paulson proposed the Emergency Economic Stabilization Act on October 3, 2008. The acceptance of which by the House of Representatives and the US Senate led to the Troubled Asset Relief Program (TARP), which reserved $700 billion dollar to buy assets and equity from

4_{Stephen Schwarzman of the Blackstone Group called the demise of Bear Stearns ‘a corporate}

(14)

financial institutions to strengthen the financial sector. This came to pass via a set of programs with different objectives. The central piece of the TARP is the Capital Purchase Program (CCP).

The CCP was designed to stimulate lending to the private sector by buying equity in Qualifying Financial Institutions (QFIs). QFIs included bank holding companies, financial holding companies, insured depository institutions and loan holding companies that are established and operating in the United States and not controlled by a foreign bank or company. In order to apply for CCP funds, the QFI had to obtain a recommendation regarding their financial health from their primary regulator, which could be the Federal Reserve Bank (Fed), the Federal Deposit Insurance Corporation (FDIC), Office of the Comptroller of the Currency (OCC) or the Office of Thrift Supervision (OTS). Based on this recommendation, the Treasury then made the decision to purchase capital or not. Successful applicants were published within two days; unsuccessful applicants were not published at all to prevent bank runs. It may be noted that this procedure was not wholly transparent.

$250 billion was reserved for this program, of which $204,9 billion was actually spent. The Treasury bought equity equal to one to five percent of the receiving bank’s risk weighted assets, in the form of senior preferred shares without voting rights. In the first five years, the dividend over these preferred shares is five percent, after that, it becomes nine percent. On top of that, the Treasury Department received warrants for fifteen percent of the investment with a strike price based on the 20 day trailing stock price. Participation was voluntary, although nine institutions were forced to join5.

Closely associated with the Capital Purchase Program is the Targeted Investment Program (TIP). Citigroup and Bank of America received additional aid to the tune of $20 billion each in November 2008 and January 2009, respectively. Both had already received $25 billion under the Capital Purchase Program, but since the CCP is notionally for healthy banks, the additional aid came under a different program. The terms of the TIP are somewhat more disadvantageous than those of the CCP. The preferred shares have an eight percent dividend. Treasury also received warrants. The TIP investments were repaid by December 2009.

(15)

Besides these programs, there is an array of other government programs designed to support the financial sector. For example, the FDIC’s Temporary Liquidity Guarantee Program (TLGP) insures debt issued by banks for a small fee. Also, the fed offers emergency lending in select cases.

The conditions of the CCP were amended by the American Recovery and Reinvestment Act (ARRA), signed into law on February 17, 2009. This act made the CCP less attractive for banks to be in, but made it easier to exit the program early. It prohibits the payment of bonuses, retention awards and incentive compensation to senior executives, although it has been noted that this can be circumvented quite easily with deferred compensation6. Later on, H.R. 1586 came into law. It levies a 90 percent tax on executive bonuses for banks that received more than 5 billion in CCP funds. Originally, recipients of TARP money were not able to exit the program in the first three years after receiving the funding, unless they issued new equity equal to or in excess of the funds provided by the Treasury Department. ARRA allowed banks to repay the funds after consulting with the appropriate Federal banking agency. No waiting period or raising of additional equity was applicable anymore.

Originally, the Troubled Asset Relief Program was estimated to cost $365 billion. The most current estimate, by Blinder and Zandi (2010), is that the program will probably cost less than $100 billion dollar, with the bank bailout component likely to turn out a profit. The fact that most of the largest banks have already paid back the government funds in 2009 contributes greatly to this fact. Blinder and Zandi (2010) conclude that although the TARP was the largest government intervention in the history of the United States, it was supremely successful both in restoring order in financial markets and stimulating the economy.

3.6 Characteristics of TARP participants

The main objective of the CCP was to get banks to start lending to the private sector again. The main objective of the TIP was to ‘stabilize the financial system by making

6_{Source: (Benjamin, Matthew and Christine Harper), “Goldman, JPMorgan Won’t Feel Effects of} Executive-salary Caps”, Bloomberg, February 5, 2009, link:

(16)

investments in institutions that are critical to the functioning of the financial system’7. In the media, it was perceived as more money for Citigroup and Bank of America to dispel rumors of a possible bankruptcy.

Although the program was open only to healthy banks, to the public eye and the media it has become known as the bailout. The opaque application procedure has contributed to this. Ng, Vasvari and Wittenberg (2010) find that banks with a higher profitability, lower ratio of non-performing to total loans and a lower book-to-market equity ratio were more likely to participate in the program, thus suggesting that the participating banks were healthier. They also observe that banks with stronger capital ratios and higher liquidity were less likely to participate in the program, suggesting that participation in the CCP was reserved for banks with stronger fundamentals that experienced lower levels of regulatory capital and liquidity. This finding is supported by Bayazitova and Shivdasani (2009), who also conclude that TARP aid was given to banks most likely to pose systemic risk. They observe that the value enhancement from TARP capital arises out of lowered costs of financial distress and not from the possible certification effect of TARP.

Duchin and Sosyura (2009) note another interesting relation: consistent with Schleifer and Vishny (1994), they find that a bank’s political ties play a significant role in TARP distribution. Connections to Congressmen on finance committees and representation at the Federal Reserve via board members positively influence the likelihood of receiving TARP capital. The influence is stronger for underperforming banks. This is supported by Li (2010).

Veronesi and Zingales (2008) valued the first 10 transactions of the CCP, concluding that the Treasury Department paid $125 billion for claims worth only $89-112 billion. The Congressional Oversight Panel’s evaluation report on February 6, 2009 corroborates this, stating that over all transactions in 2008, Treasury paid $254 billion for assets worth only $176 billion. Veronesi and Zingales do conclude that, from an economic point of view, the plan was a success, as it created value, if only through a significant wealth redistribution from tax payers to bond holders.

(17)

There have been doubtful reports in the media as to the efficacy of the program in getting the banks to lend again8. Li (2010) concludes that overall, the CCP was successful in stimulating bank lending to the private sector. Banks with a below median Tier 1 ratio increased bank loans by 6,41 percent, annualized, from the third quarter of 2008 through the second quarter of 2009. Taliaferro (2009), on the other hand, argues that most of the TARP funds were used to fill capital holes. He finds that participating banks used only thirteen cents on every program dollar on increased lending and 60 cents on shoring up their capital ratios.

It is observed by Ng, Vasvari and Wittenberg (2010) that participating banks were more profitable and experienced a smaller increase in non-performing loans. They were also less likely to delist. Consistent with Taliaferro (2009), no differences in lending activity between participating and non-participating banks are found.

3.7 Repayments of TARP money

For the first five years of the CCP program, the dividend on the preferred stock is five percent, which is lower than the average bank’s cost of capital9. It would therefore be sensible from an economic point of view to remain in the TARP at least for five years since the date of participation. There are at least three reasons why banks nonetheless decided to exit the program:

1. The executive payment cap. Under ARRA, banks cannot awards bonuses, retention awards or incentive payments. H.R. 1586 levies a 90 percent tax on executive income above $250,000 for institutions receiving more than $5 billion. This makes it harder for the TARP participants to attract and keep personnel. This is reflected in the stock market: on the day H.R. 1586 was passed, the stock of TARP recipients experienced an average excess return of eight percent. Banks that did not participate experienced a positive excess return (Kim, 2010).

2. TARP has become something of a political instrument. Since the inception of the program, TARP recipients have faced more and more conditions. They have to put off evictions, modify mortgages for distressed homeowners, let shareholders vote on

8_{For example, see (Appelbaum, Binyamin.), “Despite Federal Aid, Many Banks Fail to Revive Lending”,} Washington Post, Feb 3, 2009. link:

http://www.washingtonpost.com/wp-dyn/content/article/2009/02/02/AR2009020203338.html

9_{Source: Aswath Damodaran, Cost of capital by sector 2010, see:}

(18)

executive pay packages, slash dividends, cancel employee training and morale-building exercises and withdraw job offers to foreign citizens10. Also, banks experience uncertainty about even more additional legislation.

3. The perception about the TARP has been shifted. Although it is a program to boost lending to the private sector, it is now perceived as government help to weak banks, thus becoming somewhat of a “scarlet letter”. The legislation listed above contributes to this stigma. Several CEOs cited this as the reason to exit the program11.

Before exiting the program, financial institutions must have the approval of the Treasury, the Federal Reserve and the FDIC. This is based on ‘… [The] institutions’ overall soundness, capital adequacy, and ability to lend, including confirming that [institutions] have a comprehensive internal capital assessment process’. Specifically, ‘[They] must be able to demonstrate [their] financial strength by issuing senior unsecured debt for a term greater than five years not backed by FDIC guarantees, in amounts sufficient to demonstrate a capacity to meet funding needs independent of government guarantees’ 12_{. The FDIC guarantees refer to the Temporary Liquidity}

Guarantee Program (TLGP).

So while a bank’s financial condition may not in itself be a reason for exiting the program, it is at least instrumental to it. Wilson and Wu (2010) apply a probit regression to explain TARP exit, and find that banks that raised private capital in 2009, had better earnings performance in 2008 and have fewer problem loans are significantly more likely to have exited the TARP early. This is supported by Ng, Vasvari and Wittenberg (2010).

Wilson and Wu (2010) find only limited evidence that TARP repayment was associated with CEO pay restrictions. They do note that the first eight participating banks (that were forced into the program) ‘all at least partly exited the government’s embrace’.

10_{(Labaton, Stephen.), “Some banks, citing strings, want to return aid”, New York Times, March 10, 2009,} link: http://www.nytimes.com/2009/03/11/business/economy/11bailout.html?hp

11_{Jamie Dimon of JP Morgan Chase and Joseph DePaolo of Signature Bank, see: (Hester, Elizabeth.),} “Dimon Says He’s Eager to Repay ‘Scarlet Letter’ TARP (Update3)” Bloomberg, April 16, 2009, link:

http://www.bloomberg.com/apps/news?pid=newsarchive&sid=az0FiElfwtoM and “CEO Who Returned Funds Notes TARP's Stigma”, National Public Radio, April 1, 2009, link:

http://www.npr.org/templates/story/story.php?storyId=102618967

(19)

4. Analysis

Here, I formulate my hypotheses, list the variables used and explain the model with which the hypotheses will be tested.

4.1 Conceptual Model and hypotheses

How does executive pay relate to the speed with which banks have repaid the TARP funds? This relationship depends on the supposed effect TARP has on banks, the supposed effect executive pay has on company performance and the financial conditions during entry and before exit. Below I discuss these relationships.

The effect of TARP on banks

Of the effect TARP has on banks, it can most certainly be said that it provides banks with cheaper financing than they would have otherwise, if only for the first five years, which provides a strong incentive to pay back the TARP funds at a later stage. On the other hand, it limits executive compensation, imposes additional legislation and imbues the participating firms with a social stigma. Whether an early payback is economically sound depends on whether the negative value effect of the executive compensation cap, additional legislation and social stigma exceeds the positive value effect due to cheap financing. Bayazitova and Shivdasani (2009) find evidence that this is indeed the case. After announcing withdrawal from the program, banks experience positive returns. I therefore conclude that withdrawing from TARP, even in the first five years in which they enjoy cheap financing, is an economically sound decision. If this were not the case, an early repayment of the TARP funds would be a sign of agency problems.

The effect of performance-based pay

(20)

is high during recessions (Li, 2007). I therefore hypothesize that high performance-based compensation before the crisis has ultimately led to worse performance while high performance-based compensation after the crisis led to better performance.

Financial condition of the TARP participants

TARP recipients are for the most part well-functioning companies with low capital ratios and liquidity, as documented by Ng, Vasvari and Wittenberg (2010).

In order to gain approval from the regulatory agencies, the banks need to have proper capital adequacy and liquidity and must be able to attract debt funding. Therefore, I hypothesize that higher capital adequacy, liquidity and net operating profit all shorten the time a bank stays in the TARP program.

Figure 4.1 Conceptual Model

(21)

start of the TARP program. In figure 3.1, this relation in shown. In the next section, the various supposed relationships between the variables and time to repayment are discussed per category.

4.2 Hypotheses

I make a distinction between the CEO compensation in 2008 and in 2009. The underlying rationale is that the performance-based compensation in 2008 stimulated CEOs to take risk, while the performance-based compensation in 2009 stimulated CEOs to build up the firm’s financial strength. So, concordantly, the performance-based pay in 2008 is expected to correlate positively with time to repayment, and the performance-based pay in 2009 is expected to correlate negatively with time to repayment. Long term incentive compensation, which aligns the interests of manager and owner better than short term incentives (Wetphal and Zajac, 1994), is expected to correlate negatively with time to repayment for 2008 and 2009.

I test the effect of the total compensation as a percentage of the firm’s equity, the effect of the performance-based compensation as a percentage of total compensation, and the effect of long-term incentive compensation as a percentage of performance-based compensation.

Hypothesis 1a. The total CEO compensation as a percentage of total equity in 2008 is positively related to the time to repayment.

Hypothesis 1b. The performance-based compensation as a percentage of the total CEO compensation in 2008 is positively related to the time to repayment.

Hypothesis 1c. The long-term incentive compensation as a percentage of the total performance-based compensation in 2008 is negatively related to the time to repayment.

Hypothesis 1d. The total CEO compensation as a percentage of total equity in 2009 is negatively related to the time to repayment.

(22)

Hypothesis 1f. The long-term incentive compensation as a percentage of the total performance-based compensation in 2009 is negatively related to the time to repayment.

The return on assets of larger financial institutions is less subject to risk, because the assets are more diversified. That is why larger financial institutions typically have a lower capital ratio (Demsetz and Strahan, 1997). I expect larger firms to be able to recapitalize faster on account of their more diversified asset base, thus leading to less time to repayment. Wilson and Wu (2010) find that larger banks have a higher likelihood of exiting TARP.

Capital adequacy is an important condition which must be met before a bank is allowed to exit TARP by the federal regulators. Therefore, I expect a positive relation to exist between the repayment speed and capital adequacy, as measured by total risk-based capital. Wilson and Wu (2010) find that banks that are higher capitalized are more likely to exit the program.

On the long term, high profitability builds capital reserves (Flannery and Rangan, 2003). Therefore, higher profitability is associated with a shorter time to repayment. Lack of liquidity was one of the problems the Treasury sought to relieve with the TARP. I therefore expect that a firm’s liquidity, as measured by noncurrent assets to total assets, is necessary to exiting the program.

A stock listing greatly increases a bank’s ease of funding. Both directly (through selling stock) and indirectly (the information the bank is required to divulge make it easier for creditors to provide financing. I expect listed companies to take less time to repayment than unlisted firms.

Hypothesis 2a. The size of the firm is negatively related to time to repayment Hypothesis 2b. The capital ratios of a firm are negatively related to time to

repayment.

Hypothesis 2c. The profitability of a firm is negatively related to time to repayment. Hypothesis 2d. The liquidity of a firm is negatively related to time to repayment Hypothesis 2e. Firms that have a stock listing have a shorter time to repayment than

(23)

4.3 Data and variables

I construct a sample of 164 TARP recipients. These include all 82 banks which have repaid the TARP funds partly or in whole and 82 banks which were randomly drawn from the group that as of yet has not repaid any funds.

Of each company, data on TARP funds and repayment, 2008 10-K and 10-Q accounting data, CEO compensation data and data on stock listing are used. The data on TARP distribution and repay is obtained from Propublica.org13_{. 2008}

accounting data is gathered from the Federal Deposit Insurance Corporation’s (FDIC) institution directory14. The CEO compensation data is collected from the Securities and Exchange Commission’s (SEC) DEF 14A forms15. Of 101 TARP participants, these forms were available. The other banks are private. Finally, stock listing data comes from NASDAQ16_{. The variables are listed in table 3.1.}

13_{http://bailout.propublica.org/initiatives/2-emergency-economic-stabilization-act} 14_{http://www2.fdic.gov/idasp/main.asp}

(24)

Table 4.1 Variable definitions

Category Name Definition

TTR Time to repayment, defined as the (estimated) number of days the bank has participated in the TARP, see section 4.4

1. TARP

DIP Dummy indicator, 0 denotes voluntary participation, 1 denotes involuntary participation

TC8 Total CEO compensation as a percentage of total firm equity in 2008 times 1.000

PP8 Percentage of performance-based compensation (includes cash bonuses, restricted stock awards, stock option awards, deferred incentive

compensation incl. pensions) over total compensation in 2008

PL8 Percentage of long term incentive compensation (includes restricted stock awards and deferred incentive compensation incl. pensions) over

performance-based compensation in 200818

TC9 Total CEO compensation as a percentage of total firm equity in 2009 times 1.000

PP9 Percentage of performance-based compensation over total compensation in 2009

2. CEO Compensation17

PL9 Percentage of long term incentive compensation over performance-based compensation in 2009

LTA Logarithm of the average total assets of the firm over Q2 2007 – Q2 2008, indicates size

RBC Total risk-based capital as a percentage of risk-weighted assets in Q2 2008, indicates capital adequacy

NOI Net operating income as a percentage of average assets in Q2 2008, indicates profitability

NCA Noncurrent assets as a percent of total assets in Q2 2008. Noncurrent assets are defined as assets that are past due 90 days or more plus assets placed in nonaccrual status plus other real estate owned (excluding direct and indirect investments in real estate). Indicates liquidity

DNY Dummy indicator, 1 denotes listing on the New York Stock Exchange (NYSE), 0 otherwise.

3. Financial condition and funding

DNA Dummy indicator, 1 denotes listing on the NASDAQ, 0 otherwise

4.4 Time to repayment

The payout speed is calculated as the time between receiving the TARP funds and paying them back. As of July 31, 2010, only 83 of the 707 participating institutions

17_{The details of the CEO compensation per company can be found in appendix B.}

(25)

have paid the money back, partly or in whole. This presents a problem, as it is absolutely unknown when the other institutions will pay their money back. I handle this problem as follows: under the assumption that every participating institution faces the same broad economic conditions, it can be argued that every institution will pay back the TARP funds in about the same time, conditional upon their general financial health. This is expressed in formula (3.1).

(4.1)





















=

⋅

<











 −

⋅

=

0

1

0

1

1 )

(

PB

DP

PB

DP

PB

DP

TTR

χ

With TTR being Time to Repayment, DP being the number of Days Passed, PB being the percentage of funds Paid Back, and χ is a penalty parameter for institutions that have not paid back any TARP funds.

For firms that have paid back the TARP funds in full, the time to repayment is just the number of days they have been in the program. For firms that have not paid back any funds, the time to repayment is the number of days they have been in the program thus far times penalty parameter χ, which can be one or more. Would this parameter be one, then one assumes that all the firms that have not paid back the funds do so today. If the parameter is, for example, 1,5, then one assumes that the firms will pay back the money after 1,5 times the number of days they have been in the program thus far. Firms that have paid back the TARP funds in part are estimated to pay back the remaining funds with exactly the same speed as they have paid them back as of now.

(26)

4.5 Model

The hypotheses will first be tested using bivariate tests: Pearson correlation tests for the continuous variables, and independent T-tests for the dummy variables. Then a linear regression model will be tested using ordinary least squares. In this model, all variables will be used to explain the time to repayment. In order to obtain robust results, the bivariate tests and the linear regression model will be tested using both the whole sample and only the sample of firms that have paid back the TARP funds in full. To test the robustness of the penalty parameter χ, the correlations and multivariate models are estimated with χ=1,5 (firms that have not paid anything back need 50 percent more days to pay it back), χ=2 (firms that have not paid anything back need 100 percent more days to pay it back), χ=2,25 (firms that have not paid anything back need 125 percent more days to pay it back).

(27)

5. Results

For every hypothesis category, I will give descriptive statistics, bivariate correlations and the outcomes of the multivariate ordinary least squares regression. The results displayed in section 4 are tested with χ=2. No significant differences between results arising out of different values of χ are found in the bivariate tests. I start with a brief discussion of the time to payment variable.

5.1 Time to repayment

In table 5.1, the descriptive statistics for the time to repayment variable can be seen. Centra Financial Holdings participated in the program for just 74 days. SunTrust is currently the bank that is longest in the program, at 628 days19_.

Table 5.1 Time to repayment descriptive statistics

Minimum Maximum Mean St. dev.

No. of days in TARP for banks that paid back

74 572 279,67 137,39

No. of days in TARP for all banks 74 628 430,24 168,09 TTR(χ=1,5) 74 944 548,05 296,12 TTR(χ=2) 74 1258 685,92 428,08 TTR(χ=2) 74 1415 754,86 495,25 5.2 CEO compensation

The CEO’s average total compensation in 2008 is 0,72 percent of the bank’s equity. The performance-based percentage varies between 4 and 100 percent, with an average of 48 percent. Of this, on average 52 percent is based on long-term.

In 2009, the average total compensation as a percentage of equity in 2009 is both higher and more widely distributed. This is a reflection of the decrease in equity values, as the CEOs did not get higher pay in absolute terms. Strikingly, the percentage of performance-based compensation was lower (on average 29 percent) but of this, a higher percentage was long-term based (73 percent). The banks that paid back the TARP funds have lower total compensation in 2008, but a higher

(28)

percentage was performance-based (62 percent). Of the performance-based compensation, a higher percentage was long-term incentive (60 percent) than the banks that did not pay back. Broadly, the same picture can be sketched for 2009.

Table 5.2 Descriptive statistics for CEO compensation

All banks Banks that paid back

Sample

Name Mean St.Dev Mean St.Dev

TC8 7,17 11,45 4,82 5,74 PP8 0,48 0,27 0,62 0,23 PL8 0,52 0,38 0,6 0,32 TC9 7,83 14,46 7,18 15,91 PP9 0,29 0,98 0,54 0,38 PL9 0,73 0,35 0,75 0,31

Total compensation in 2008 correlates significantly and positively with time to repayment, which confirms hypothesis 1a. Contrary to expectation, the percentage of performance-based pay correlates negatively with time to repayment. Hypothesis 1b is rejected. Of the performance-based pay, the percentage of long-term incentive pay shows a negative correlation with time to repayment, confirming hypothesis 1c. Both total compensation in 2009 and the percentage of long-term incentive pay do not correlate significantly with time to repayment, whilst performance-based pay does correlate significantly and negatively, Hypotheses 1d. and 1f. are rejected, while hypothesis 1e is confirmed.

(29)

Table 5.3 Pearson correlation coefficients for CEO compensation and time to repayment

Sample Name

TC8 0,23** -0,04 PP8 -0,53*** -0,05 PL8 -0,24** 0,06 TC9 0,1 0,08 PP9 -0,20** -0,24* PL9 -0,07 0,31**

*** (**,*) indicates two-tailed significance at the 1% (5%, 10%) level

5.3 Financial condition and funding

In table 5.4, descriptive statistics pertaining to the financial characteristics of the bank in the sample can be seen. What is immediately striking, though not unexpected, is that banks that paid back have a better financial health than all the banks as a whole.

Table 5.4 Descriptive statistics for financial characteristics

Sample

Name Mean St.Dev Mean St.Dev

(30)

Table 5.5 Pearson correlation coefficients for financial characteristics and time to repayment

Sample Name

LTA -0,32*** -0,05

RBC -0,16** 0,00

NOI -0,25*** -0,27**

NCA 0,28*** 0,08

For the time to repayment of the sample containing all banks, there are significant negative correlations with size, total risk-based capital ratio, and net interest income to earning assets, confirming hypotheses 2a, 2b and 2c. There are significant positive correlations for the indicator of liquidity, noncurrent assets to total assets, confirming hypothesis 2d. Interestingly, only net operating income to earning assets correlates significantly with time to repayment for the sample of banks that already paid back.

Table 5.6 T-test of stock listing on time to repayment

Sample Variable 0 1 Dif. St. Error t

NYSE (1) vs. NASDAQ (0) 697,77 476,44 221,32 112,08 -1,98** NYSE (1) vs. no listing (0) 860,20 476,44 383,39 98,65 3,89*** All banks NASDAQ (1) vs. no listing (0) 860,20 697,77 162,43 70,07 2,32** NYSE (1) vs. NASDAQ (0) 266,63 273,94 7,31 40,35 0,18 NYSE (1) vs. no listing (0) 319,20 273,94 45,26 45,68 0,99 Banks that paid back NASDAQ (1) vs. no listing (0) 319,20 266,63 52,57 43,37 1,21

(31)

hypothesis 3d. No significant results can be found for the sample of firms that paid back all.

A large number of hypotheses can be confirmed using bivariate tests. Apart from net operating income over earning assets and percentage of performance-based compensation, no explanatory variable discriminated significantly within the sample of banks that paid back. This may be because the group is at this moment in time (July 31, 2010) too homogeneous with respect to the explanatory variables.

In section 5.4, the results of the linear regression model can be found. It may be noted that the concomitant variation between time to repayment and the explanatory variables seems to arise for a large part out of the difference between companies that have as of now paid back the TARP funds and the firms that have not. To research this, I also construct a binary logistic regression model, to be found in section 5.6.

5.4 Linear regression model

Now I construct a multivariate linear model, using an ordinary least squares regression. Because only one explanatory variable discriminated significantly within the sample of banks that paid back, the regression will only be performed on the whole sample.

The ANOVA test of the linear regression model is significant, thus rejecting the null hypothesis that all coefficients are equal to zero. The R2_{is 0,365, indicating that the}

independent variables explain 36,5 percent of the variation in time to repayment. The values of the coefficients20 are listed in table 5.7.

(32)

Table 5.7 Variable coefficients of the linear regression model with time to repayment as the dependent variable

Name Variable B St. Error T

(Constant)

1784,42 316,44 5,64***

TC8 Total CEO compensation over total equity in

2008 times 1.000 1,10 3,30 0,33

PP8 Percentage of performance-based CEO pay

in 2008 -681,33 185,56 3,67***

PL8 Percentage of long-term incentive

performance-based pay in 2008 -129,42 97,66 1,36

PP9 Percentage of performance-based CEO pay

in 2009 -27,93 38,41 0,73

LTA Logarithm of average total assets _-25,66 _20,76 _1,29

RBC Total risk-based capital ratio

-29,85 9,76 3,06***

NOI Net operating income

-73,49 28,41 2,59***

NCA Noncurrent assets to total assets

67,89 23,50 2,89***

DNY NYSE listing _42,42 _146,69 _0,29

DNA NASDAQ listing

-124,42 72,90 1,71*

No notable differences in parameters and parameter significance arise out using a different penalty parameter χ for firms that as of yet have not paid back all. In the multivariate linear regression model, percentage of performance-based CEO compensation in 2008, total risk-based capital ratio, net operating income and noncurrent assets to total assets are significant on a five percent level. NASDAQ listing is significant on a ten percent level.

(33)

that the percentage of performance-based CEO compensation is a major influence on time to repayment.

From the multivariate model, it appears that a NASDAQ listing shortens the time to repayment, significant on a ten percent level. A NYSE listing does not have any significant effect. This casts doubt over hypothesis 2d., more on which will be said further on.

5.5 Binary logistic regression model

The binary logistic regression model is used to explain the binary event (0 = the firm has not paid back the TARP funds in full, 1 = the firm has paid back the TARP funds in full) with the same explanatory variables as the linear regression model. The binary logistic regression model will, naturally, be done on the whole sample.

(34)

Table 5.8 Variable coefficients of the binary logistic regression model with payback (0,1) as the dependent variable.

Name Variable B St.Dev Wald Exp

(Constant)

-10,65 4,77 4,98 0,00

TC8 Total CEO compensation over total

equity in 2008 times 1.000 0,01 0,05 0,05 1,012

PP8 Percentage of performance-based

CEO pay in 2008 5,70 2,06 7,69*** 298,98

PL8 Percentage of long-term incentive

performance-based pay in 2008 0,58 0,88 0,43 1,784

PP9 Percentage of performance-based

CEO pay in 2009 0,70 0,82 0,74 2,022

LTA Logarithm of average total assets

0,33 0,29 1,27 1,386

RBC Total risk-based capital ratio

0,26 0,13 4,28** 1,299

NOI Net operating income

0,59 0,30 3,79* 1,801

NCA Noncurrent assets to total assets

-1,25 0,52 5,80*** 0,286

DNY NYSE listing

-0,78 1,60 0,24 0,460

DNA NASDAQ listing

1,17 1,11 1,11 3,221

In the binary logistic regression model, as in the linear model, percentage of performance-based CEO compensation in 2008, total risk-based capital ratio, net operating income and noncurrent assets to total assets are significant at the five percent level. This offers support for the general idea that the speed with which firms have repaid TARP depends on their financial health and performance at the start of the program. It also confirms that the percentage of the CEO’s compensation that is performance-based in 2008 actually contributes to a speedy repayment. In fact, holding all the other variables constant, a one percentage increase in performance-based CEO pay makes it almost twenty percent more likely that the firm has paid off the TARP funds by July 31, 2010. Although this effect is very strong, there are still prominent examples of firms that, although over 80 percent of the CEO compensation is performance-based, still have not paid back. These include Citigroup and SunTrust.

(35)

significant result for NASDAQ in the linear regression. I cannot state that listing has any effect on repayment when the other variables are taken into account. The same can be said of company size.

The outcomes of both models taken together, it can be said with some confidence that the speed with which banks pay back the TARP money is strongly dependent on the percentage of performance-based CEO compensation before the crisis. Also notable is the fact that the total size of the compensation package and the percentage of long-term incentive compensation do not seem to matter. Concluding, performance-based pay may have increased the risk-taking within banks before the crisis (Cheng, Hong and Scheinkman, 2009), but it also increases performance after. It cannot be ascertained that this is due to increased risk-taking after the crisis.

Compensation figures from 2009 have no effect in a multivariate model. The percentage of performance-based CEO compensation in 2009 does correlate negatively with time to repayment in a bivariate analysis, but this effect is seemingly not strong enough to turn up significant in a multivariate analysis. This may be due to the fact that the compensation packages of CEOs were altered as a result of TARP or because of public pressure. For example, Vikram Pandit of Citigroup reduced his salary to $1,- per year in February 200921.

21_{(Farrell, Gregg.), “Citigroup Chief Awarded $10.82m”, Financial Times, March 17, 2009, link:}

(36)

6. Conclusion

As a consequence of their business practice, banks are almost always short on liquidity. During the credit crisis, liquidity all but dried up. As banks are very sensitive to systemic risk, the credit crisis affected all banks, although badly managed banks were disproportionally affected. This brought the banking system to the brink of collapse.

The US government stepped in with the Troubled Asset Relief Program (TARP). Under this program, the Treasury buys preferential shares in healthy financial institutions. The first three years, the dividend is five percent, after that, nine percent. The first three years, during which TARP arguably presented a cheap form of financing for the banks, are not over yet, but banks have started paying back the money and exiting the program.

The reasons for this are that 1) TARP limits executive payment, thus making it harder for banks to attract and retain personnel, 2) TARP imposes additional legislation on banks, which hinders business and brings with it uncertainty about still more legislation and 3) TARP brings a negative social stigma to participating banks. These disadvantages outweigh the advantage of cheap financing, as evidenced by the fact that the stock goes up upon the announcement of paying back the TARP funds (Bayazitova and Shivdasani, 2009).

This research investigates the link between CEO compensation and the time it takes banks to pay back the TARP funds, specifically the percentage of CEO compensation that is performance-based in some form or another. Performance-based compensation aims to reduce agency costs, but may increase them as well (Guay, 1999). I find that, controlling for indicators of financial condition, size and listing, a higher percentage of performance-based pay in 2008 (before the crisis) leads to a shorter time to repayment in both a linear model and a binary logistic model. So, irrespective of their financial condition, size and listing do banks pay back the TARP funds quicker if the CEO is rewarded with more performance-based compensation.

(37)

CEOs based on performance leads to better performance. It must however be noted that the sample only included TARP recipients, which are on average well-functioning banks (Ng, Vasvari and Wittenberg, 2010). So the results can only be generalized to well-functioning companies, and offer no reflection on the banks that collapsed before the program was initiated. The results lead one to suggest that if a corporate governance framework to prevent irresponsible risk-taking is in place (Kirkpatrick, 2009), a company performs better if the CEO is paid in terms of performance.

A limitation of this research worth mentioning is that not all information about the repayment of TARP is available, simply because not all banks have paid back the funds in full yet. This has been accounted for by estimating both a linear model and a binary logistic model, which both lead to the conclusion above. Still, replicating this research at a later stage might lead to slightly different insights.

(38)

7. References

Barr, R., L. Seiford, and T. Siems, “Forecasting Bank Failure: A Non-Parametric Approach”, Recherches Economiques de Louvain, 60 (1994): 411-29. Bayazitova, Dinara, and Anil Shivdasani, (2009), “Assessing TARP,” SSRN Working

Paper, accessed online of July 22, 2010, at

http://ssrn.com/abstract=1461884.

Bebchuk, Lucian, Jesse M. Fried & David I. Walker (2002), Managerial Power and Rent Extraction in the Design of Executive Compensation, University of Chicago law Review 69 (2002)

Bebchuk, Lucian and Itay Goldstein, (2008), “Self-Fulfilling Credit Market Freezes,” Harvard, John M. Olin Center for Law, Economics, and Business

Discussion Paper No. 623.

Bebchuck, Lucaion and Holger Spamann (2009), “Regulating Bankers’ Pay”, Harvard Law School Discussion Paper 641.

Berger, Allen N., Seth D. Bonime, Daniel M. Covitz, and Diana Hancock (2000), “Why Are Bank Profits So Persistent? The Roles of Product Market Competition, Informational Opacity, and Regional/Macroeconomic Shocks.” Journal of Banking and Finance

Berger, A. N., R. J. Herring, and G. P. Szegö (1995). “The Role of Capital in Financial Institutions,” Journal of Banking and Finance 19, 257-276.

Bernardo, Antonio E. and Ivo Welch, 2004, Financial Market Runs, Quarterly Journal of Economics, 119, 135-158.

Bertrand, Marianne & Mullainathan, S., Are CEOs Rewarded for Luck? The Ones without Principals Are, Quarterly Journal of Economics 116 (2001): 901– 932.

Caprio, Gerard and Patrick Honohan. 2008. “Banking Crises.” IIIS Discussion Paper No. 242.

Cheng, Ing-Haw, Harrison G. Hong, and Jose A. Scheinkman, (2009), “Yesterday’s heroes: Compensation and creative risk-taking”, unpublished working paper, Princeton University.

Demsetz, Rebecca S. and Philip Strahan (1997), “Diversification, Size, and Risk at Bank Holding Companies,” with Philip Strahan, Journal of Money, Credit, and Banking, Vol. 29, No. 3, August 1997.

Diamond, Douglas W. and Philip H.Dybvig, (1983), "Bank runs, deposit insurance, and liquidity" (PDF). J Pol Econ 91 (3): 401–19.

Diamond, Douglas W., and Raghuram G. Rajan, (2001), Liquidity risk, liquidity creation, and financial fragility: A theory of banking, Journal of Political Economy 109, 287-327.

Diamond, Douglas W. And Raghuram G. Rajam (2002), “Bank Bailouts and

Aggregate Liquidity”, The American Economic Review, Vol. 92, Nr. 2, pp. 38-41

Diamond, Douglas W. And Raghuram G. Rajam (2005), “Liquidity Shortages and Banking Crises”, The Journal of Finance, Vol. 60, Nr. 2 (April 2005), pp. 615-647

Duchin, Ran, and Denis Sosyura, (2009), “TARP Investments: Financials and Politics,” SSRN Working Paper, accessed online of July 22, 2010, at

Fahlenbrach, Rudiger and Rene Stulz (2009): ―Bank CEO Incentives and the Credit Crisis, Ohio State Working Paper.

Gavin, M and Hausmann, R (1996), ‘The Roots of Banking Crises: The

(39)

‘Banking Crises in Latin America’, John Hopkins University Press, pp. 27– 63.

Goodhart, Charles and Haizhou Huang (1999), “A Model of the Lender of Last Resort”, FMG, LSE Discussion Paper 313, London, U.K.

Goddard, John, Phil Molyneux, and John O.S. Wilson. (2004b) “Dynamics of Growth and Profitability in Banking.” Journal of Money, Credit, and Banking, 36:6, 1069-90.

González-Hermosillo, B (1999) ‘Determinants of Ex-ante Banking System Distress: A Macro-Micro Empirical Exploration of Some Recent Episodes’ IMF

Working paper,99/33.

Gorton, Gary and Lixin Huang, 2004, Liquidity, Efficiency and Bank Bailouts, American Economic Review, 94 (3), 455-483.

Guay, W., 1999. The sensitivity of CEO wealth to equity risk: an analysis of the magnitude and determinants. Journal of Financial Economics 53, 3–71. Hoggarth, G, Reis, R and Saporta, V (2002), ‘Costs of banking system instability:

some empirical evidence’, Journal of Banking and Finance, Vol. 26, No. 5, pages 825.55.

Tosi HL, Werner S, Katz JP, Gomez-Mejia L. (2000). How much does performance matter? A meta-analysis of CEO pay studies. Journal of Management, 26, 301–339.

Jensen, Michael C. and William H. Meckling. 1976. "Theory of the Firm: Managerial Behavior, Agency Costs, and Ownership Structure." Journal of Financial Economics 3, no. 4: October, pp 305-360.

Kim, Won Yong, (2010), “Market Reaction to Limiting Executive Compensation: Evidence from TARP Firms,” SSRN Working Paper, accessed online on July 22, 2010, at http://ssrn.com/abstract=1553394.

Li, G., (2007), “Time-varying risk aversion and asset prices”. Journal of Banking and Finance 31, 243-257.

Li, Lei, (2010), “TARP Funds Distribution and Bank Loan Growth,” SSRN Working Paper, accessed online of July 22, 2010, at

Merton, Robert K., (1968), Social Theory and Social Structure. New York: Free Press Murphy, K.J., 1999b. Executive compensation. In: Orley, A., David, C. (Eds.),

Handbook of Labor Economics, Vol. 3. North-Holland, Amsterdam. Ng, Jeffrey, Florin P. Vasvari, Regina Wittenberg-Moerman, (2010), “Were Healthy

Banks Chosen in the TARP Capital Purchase Program?” SSRN Working Paper, accessed online on July 22, 2010, at

Oldfield, George S. and Anthony M. Santomero, (1995), The place of risk

management in financial institutions, Working Paper, #95-05 Wharton Financial Institutions Center.

Reinhart, Carmen M. and Kenneth Rogoff (2008), “Banking Crises: An Equal Opportunity Menace”, NBER Working Paper

Santomero, A.M. and R.D. Watson (1977), ‘Determining Optimal Capital Standard for the Banking Industry’, Journal of Finance, 32(4) (September), 1267-82. Schleifer, Andrei and R. Vishny (1994), “Politicians and Firms”, Quarterly Journal of

Economics November 1994, pp995-1025

Taliaferro, Ryan, (2009), “How Do Banks Use Bailout Money? Optimal Capital Structure, New Equity, and the TARP.” SSRN Working Paper, accessed online on July 20, 2010, http://ssrn.com/abstract=1481256.

(40)

Veronesi, Pietro, and Luigi Zingales, (2010), “Paulson’s Gift,” forthcoming Journal of Financial Economics.

Wetphal, J.D. and E.J. Zajac (1994), “Substance and Symbolism in CEO’s long term incentive plans”, Administrative Science Quartertly, Vol 93, No. 3 pp 367-390.