Executive Compensation at Commercial Banks Before and After the Financial Crisis of Richard A. Lord Montclair State University

Transcription

1 Executive Compensation at Commercial Banks Before and After the Financial Crisis of Abstract Richard A. Lord Montclair State University Since the onset of the financial crisis in 2007, compensation of the top executives at banks has been a subject of intense interest. Many feel that the managers of these institutions are rewarded handsomely for good performance, but never punished for poor results. A second charge is that compensation structures at banks have not been altered as a result of the debacle. I show that while both bonuses and equity grants are insensitive to declines in earnings, the same does not seem the case for the effects of stock returns and revenue growth on compensation. I also find that the crisis has clearly had an appreciable impact in how performance-based pay is granted to bank executives. JEL Classification: J33, G21, G01 Key Words: Executive Compensation, Financial Institutions, Financial Crisis 1

2 1. Introduction Since the financial crisis, which began in the fall of 2007, swept first through the United States then much of the rest of the world, the executive compensation practices of financial institutions has been a burning subject. Even before the collapse, there had been long-simmering resentment over levels of executive pay in general, and in the financial industry in particular. During the crisis in late 2007 and early 2008, governments across the world stepped-in to bailout these institutions, which prompted more criticism of excessive levels of pay financed through the benevolence of tax payers. In the ensuing years there has also been a fierce debate over whether the design of pay packages created incentives for excessive risk-taking by financial institutions. I examine the relationship between firm performance measures and executive bonuses and equity grants at commercial banks. I employ data on the five highest paid officers at these institutions and concentrates on two specific executive compensation issues. First, are bonuses and equity grants to executives equally sensitive to poor performance as to good performance? Second, has the sensitivity of these pay elements to firm performance changed significantly since the crisis? In the next section I review the previous literature and present my hypotheses. The model specifications to test the hypotheses are described in the third section. Next, I describe the characteristics of the sample. The empirical results are then presented. Finally, I summarize the findings. 2. Literature Review and Hypotheses Murphy (1999) provides a classic study of general executive compensation practices. He notes a rapid growth in stock option compensation through the 1990 s. Cai and Milbourn (2010) and Lord and Saito (2010) find that for financial institutions, this trend continued until the dot.com crisis, peaking in about Since then, the use of restricted stock grants has begun to supplant option compensation. Traditionally, the other widely used form of performance-based compensation was bonuses. The most frequently used explicit basis for bonuses is profitability; earnings per share being the most common. Another widely used measure is growth in revenues. There is also evidence that bonuses are correlated with stock returns (see Hall and Liebman, 1998). In a study of executive compensation before the crisis at the largest financial institutions in the United States, Hodgson (2010) finds that the most common explicit performance measures used as the basis for bonuses are return on equity (ROE) and growth in revenue. Murphy also notes that most performance-based compensation plans have a major discretionary component. In fact, many are purely discretionary. There is a long-standing charge that executive compensation is insensitive to poor firm performance. Pay rises with good firm performance, but executives do not feel the pain when performance is poor. The most prominent critics of this general practice are Bebchuk, Fried and Walker (2002) and Bebchuk and Fried (2004). However, theoreticians in finance, such as Holstrom (1979), Smith and Stulz (1985) and Smith and Watts (1992), have long argued exactly the opposite is required; that it is necessary to provide convexity in executive pay packages. Because managers tend to hold poorly diversified portfolios, and they are individually risk- 2

3 averse, if they are exposed to downside risk to the same extent as upside payoffs they will tend to forgo investment in risky, but potentially value-increasing, projects. However, several years before the crisis, John and Qian (2003) warned that aligning the pay incentives of managers of highly levered financial institutions too closely with those of their shareholders would probably create excessive risk-taking incentives. Since the public ire over the bailouts of Wall Street firms has been so intense, politicians have begun arguing about curbing the amount and design of managerial pay. President Obama (June 17, 2009) stated excessive executive compensation unmoored from long-term performance or even reality rewarded recklessness rather than responsibility. At about the same time the Department of Treasury suggested guidelines for executive pay by financial firms receiving federal bailout funding (February 4, 2009), which mandated that any compensation above $500,000 must come as restricted stock grants. Many in congress consider this as a viable blueprint for broader legislation. There is also a widely held perception that the pay of bank executives has changed little since the beginning of the financial crisis. The appointment of Kenneth Feinberg as a Pay Czar to oversee the remuneration practices of institutions that received Troubled Asset Recovery Plan (TARP) funding, attests to the presumption that executives would not reduce their compensation nor revise the structure of the pay packages. However, there is clear evidence that annual pay for executives in the financial sector declined markedly during the crisis, though it is now moving back upward. A recent survey by the Comptroller of the State of New York (Goldman and Moore, 2011), shows that cash bonuses are half of their level in The deeper concern is that the structure of the compensation packages is not changing dramatically. In an influential study, Hodgson (2010) notes there has not been any notable shift away from short-term compensation targets to long-term goals. To examine these issues, I test two hypotheses regarding the design of the compensation packages of the five highest paid officers of commercial banks. The first is: H1: Executive bonuses and equity grants are as sensitive to poor performance as to good performance. The second is: H2: The relationship of executive bonuses and equity grants with firm performance are the same during and after the financial crisis (2007 through 2009) as they were in the period before (1992 through 2006). 3. Model Specification I test these hypotheses using separate regression models to explain bonuses and equity grants to executives in the US banking sector. The base model for bonus compensation is specified as: BONUS = α 0 + α 1 LBONUS + α 2 DCEO + α 3 CAP + α 4 LREV + β 1 STKRET + λ 1 ROE + δ 1 REVGR + ε. (1) 3

4 The dependent variable, BONUS, is the sum of annual bonuses and payments under long-term incentive plans to a bank executive. The next four terms LBONUS, DCEO, CAP and LREV are control variables. LBONUS is the lagged bonus for the same executive in the previous year. DCEO is a dummy variable set to one if this executive is the corporate CEO in this year, and set to zero otherwise. CAP is a proxy for the firm s capital structure; defined as the book-value of corporate liabilities, divided by the sum on the book-value of corporate liabilities, the book-value of preferred stock and the market value of common equity. LREV, the logarithm of revenues, is a proxy for firm size. A regression model with BONUS as the dependent variable, with only these four independent variables (detailed results not shown), explains over 76% of the variability in executive bonus grants. The three independent variables to proxy the firm performance factors that should influence (or be influenced by) executive compensation are STKRET, ROE and REVGR: STKRET is annual stock return, ROE is return on equity, and REVGR is growth in corporate revenue for the year. The ε is an error-term with the usual properties. A similar base model is specified for annual executive equity grants: EQGR = α 0 + α 1 LEQGR + α 2 DCEO + α 3 CAP + α 4 LREV + β 1 STKRET + λ 1 ROE + δ 1 REVGR + ε. (2) Here EQGR is the value of annual executive equity grants (stock options grants, at either Black- Scholes, 1973, or fair market value, plus restricted stock grants), and LEQGR is the value of equity grants for the previous year. All other variables are as defined above. A regression model with EQGR regressed on the four control variables alone (results not shown), explains about 67% of the variability in executive equity compensation. The first hypothesis is that the effects of the firm performance variables have a different impact on executive compensation when they are negative than when they are positive. To test this hypothesis, three one-zero dummy variables are created. D_STKRET is set to one when stock return is negative, and to zero otherwise. D_ROE is set to one when return on equity is negative, and zero otherwise. D_REVGR is set to one when revenue growth is negative, and zero otherwise. Then a set of three cross-product terms are created where each performance factor is multiplied by the corresponding dummy variable, N_STKRET, N_ROE and N_REVGR, respectively. The equation to test the first hypothesis for executive bonus compensation is specified as: BONUS = α 0 + α 1 LBONUS + α 2 DCEO + α 3 CAP + α 4 LREV + β 1 STKRET + β 2 D_STKRET + β 3 N_STKRET + λ 1 ROE + λ 2 D_ROE + λ 3 N_ROE + δ 1 REVGR + δ 2 D_REVGR + δ 3 N_REVGR + ε. (3) 4

5 The corresponding equation to test the first hypothesis for executive equity compensation is: EQGR = α 0 + α 1 LEQGR + α 2 DCEO + α 3 CAP + α 4 LREV + β 1 STKRET + β 2 D_STKRET + β 3 N_STKRET + λ 1 ROE + λ 2 D_ROE + λ 3 N_ROE + δ 1 REVGR + δ 2 D_REVGR + δ 3 N_REVGR + ε. (4) The second hypothesis is that changes in the firm performance variables have different effects on executive compensation after the onset of the financial crisis at the end of 2007 than they did before. To test this hypothesis, a one-zero dummy variable is created. D_2007 is set to one in the years 2007, 2008 and 2009, and is set to zero in prior years. Again, three interactive terms are created as the product of each firm performance measure with this dummy variable, T_STKRET, T_ROE and T_REVGR, respectively. The equation to test the second hypothesis for bonus compensation is then specified as: BONUS = α 0 + α 1 LBONUS + α 2 DCEO + α 3 CAP + α 4 LREV + µ 1 D_ β 1 STKRET + β 4 T_STKRET + λ 1 ROE + λ 4 T_ROE + δ 1 REVGR + δ 4 T_REVGR + ε. (5) The model to test the second hypothesis for executive equity grants is: EQGR = α 0 + α 1 LEQGR + α 2 DCEO + α 3 CAP + α 4 LREV + µ 1 D_ β 1 STKRET + β 4 T_STKRET + λ 1 ROE + λ 4 T_ROE + δ 1 REVGR + δ 4 T_REVGR + ε. (6) Each of these models is estimated using pooled cross-sectional data, drawing observations on all of the executive salaries reported for each firm, for each available year between 1992 and Because the firms in the industry are relatively homogeneous, I do not employ firm-specific or executive-specific dummy variables. I make an initial estimation of each model to identify any outliers that could seriously bias parameter estimates. All observations that have residuals with a Cook s D value higher than one and/or an R-Student value with an absolute value greater than three are eliminated. See Welsch (1980) for more details on these methods. 4. Data Sources and Characteristics The sample is composed of commercial banks with executive compensation data available on the EXECUCOMP database. This source contains comprehensive information on the pay of the five most highly paid officers for a sample of firms corresponding roughly to the S&P 1,500. The data is available from 1992 through The executive compensations variables, BONUS, LBONUS, EQGR and LEQGR and the information to create the dummy variable to identify CEOs, are all obtained from EXECUCOMP. All other variables are collected from the COMPUSTAT database. The sample is restricted to commercial banks, in the three-digit SIC code 602. Much of the popular discussion of compensation at banks is actually about large investment banking 5

6 institutions. The results from this sample will not necessarily shed any light on the behavior or performance of these firms. Summary statistics for the fourteen variables used in this study are given in Table 1. There are several interesting characteristics in the data. As expected the financial leverage (CAP) used by these banks is very high. The management compensation variables (BONUS, LBONUS, EQGR and LEQGR) and firm size (LREV) are all highly skewed. Means for the dummy variables show that 15% of the executives are CEOs, about 33% of observations have negative stock returns, almost 29% have negative annual revenue growth, less than 6% have negative ROE, and about 16% of observations are during the crisis and post-crisis years 2007 through TABLE 1 SAMPLE DIAGNOSTIC STATISTICS Observations Median Mean Std Dev Minimum Maximum BONUS 9, , , LBONUS 7, , , EQGR 7, , , , LEQGR 6, , , , DCEO 9, CAP 8, LREV 8, , , , STKRET 8, ROE 8, REVGR 8, D_STKRET 9, D_ROE 9, D_REVGR 9, D_2007 9, BONUS Annual Executive Bonus Compensation (Including Payments under Long-Term Incentive Plans). LBONUS - Lagged Annual Executive Bonus Compensation. EQGR - Annual Executive Equity Grants (Options and Restricted Stocks). LEQGR Lagged Annual Executive Equity Grants. DCEO One/Zero Dummy Variable set to One if the Executive is the CEO. CAP Capital Structure (Ratio of Liabilities to the Market Value of Assets). LREV Logarithm of Revenues. STKRET Annual Stock Return. ROE Return on Equity. REVGR Percentage Growth in Revenues. D_STKRET One/Zero Dummy Variable set to One if Stock Returns are Negative. D_ROE One/Zero Dummy Variable set to One if Return on Equity is Negative. D_REVGR One/Zero Dummy Variable set to One if Revenue Growth is Negative. D_2007 One/Zero Dummy Variable set to One if the Observation is in the Fiscal Year 2007, 2008 or Empirical Results Results for the three models that explain executive bonuses as a function of firm performance are given in Table 2. The estimates for the base model, equation 1, are shown in the first column. All of the control variables are statistically significant. As expected, current bonuses are positively correlated with bonus for the prior year, and with firm size. Not surprisingly, CEOs receive 6

7 higher bonuses than other executives. Bonuses are negatively correlated with the use of financial leverage. As anticipated, bonuses are positively related with stock returns and growth in revenue. Surprisingly, they are negatively correlated with ROE. Earnings are usually the fundamental explicit determinant of bonus payments, so a positive relationship was anticipated. However, the results in the other two columns shed some light on possible reasons for this negative sign. TABLE 2 RELATIONSHIP BETWEEN EXECUTIVE BONUS COMPENSATION AND FIRM PERFORMANCE MEASURES Equation 1 Equation 3 Equation 5 Coefficient Estimate T-Value Estimate T-Value Estimate T-Value Intercept α *** * * LBONUS α *** *** *** DCEO α *** *** *** CAP α *** ** *** LREV α *** *** *** D_2006 µ STKRET β *** *** D_STKRET β *** N_STKRET β *** T_STKRET β ROE λ *** *** *** D_ROE λ *** N_ROE λ 3-1, *** T_ROE λ *** REVGR δ *** *** *** D_REVGR δ * N_REVGR δ * T_REVGR δ *** Obs 7,037 7,037 7,035 Adj R F-Value 3,272.70*** 1,793.72*** 2,128.40*** *** Significant at 99% Level, ** Significant at 95% Level, * Significant at 90% Level, BONUS Annual Executive Bonus Compensation (Including Payments under Long-Term Incentive Plans). LBONUS - Lagged Annual Executive Bonus Compensation. DCEO One/Zero Dummy Variable set to One if the Executive is the CEO. CAP Capital Structure (Ratio of Liabilities to the Market Value of Assets). LREV Logarithm of Revenues. STKRET Annual Stock Return. ROE Return on Equity. REVGR Percentage Growth in Revenues. D_STKRET - One/Zero Dummy Variable set to One if Stock Returns are Negative. N_STKRET - Interaction Between STKRET and D_STKRET. D_ROE - One/Zero Dummy Variable set to One if Return on Equity is Negative. N_ROE - Interaction Between ROE and D_ROE. D_REVGR - One/Zero Dummy Variable set to One if Revenue Growth is Negative. N_REVGR - Interaction Between REVGR and D_ REVGR. D_ One/Zero Dummy Variable set to One if the Observation is in the Fiscal Year 2007, 2008 or T_STKRET - Interaction Between STKRET and D_2007. T_ROE - Interaction Between ROE and D_2007. T_REVGR - Interaction Between REVGR and D_

8 The results for equation 3 to test the first hypothesis for bonus payments are shown in the second column of Table 2. This specification examines whether bonuses are insensitive to poor firm performance. Parameter estimates for the control variables are similar to those for the base model. Here β 1 on stock return is no longer statistically significant. But, β 2 is negative and significant, and β 3 is positive and significant. Collectively these results suggest that bonuses are little affected by positive stock return, but they decline dramatically when stock prices fall. Leone, Wu and Zimmerman (2006) study a broad sample of firms and find that cash compensation is more sensitive to stock price declines than increases. In equation 3, λ 1 on ROE is positive as expected. The parameter estimate on the interactive term, λ 3, is significantly negative, and of roughly the same magnitude as for λ 1. This means that when earnings are positive, bonuses rise with ROE, but when earnings are negative bonuses are flat. This is exactly the relationship between earnings and bonuses described by Murphy (1999). Interestingly, λ 2 on the dummy variable is positive and significant, suggesting managers of firms with negative ROE have somewhat higher bonuses than those of profitable firms, all else the same. The parameter δ 1 is again positive, suggesting that bonuses rise with increases in firm revenues. Both δ 2 and δ 3 would only pass a confidence test at the 90% level. Since they are both negative, they tell somewhat conflicting stories. The negative sign on the dummy variable, δ 2, sensibly suggests that bonuses are lower when firm size declines. On the other hand, δ 3 on the interactive term, is opposite the sign on δ 1 and is of a greater magnitude, which might suggest that the correlation becomes negative when revenue declines. But, since the values of revenue growth are, by definition negative in this range, this implies that as firm size declines bonuses rise. Results for equation 5 to check the second hypothesis, whether executive bonus compensation changed during the financial crisis and post-crisis years of 2007 through 2009, are shown in the third column of Table 2. Again, the parameter estimates on the control variables are similar to those in the base model. The parameter on the dummy variable for the crisis period, µ 1, is not statistically significant. The coefficient β 1 is positive and significant, meaning bonuses rise with increasing and fall with declining stock returns. The parameter, β 4 is not significant, which means this relationship is no different during and after the crisis from the period before. In this model, λ 1 is positive and significant, which implies a positive relationship between ROE and bonuses before The parameter on the cross-product term, λ 4, is negative, significant and of roughly the same magnitude as λ 1, which implies that the linear relationship disappears during and after the crisis. Clearly, the incidence of negative measures of ROE was much higher in this period; overall ROE is negative for only about 5.50% of the observations, but in the three crisis and post-crisis years almost 27% were negative. The coefficient δ 1 on revenue growth is again positive. But, in the crisis period the significant negative magnitude of δ 4 might suggest a negative relationship between revenue growth and bonuses. The empirical results for the three models to explain the relationship between executive equity grants and firm performance are given in Table 3. The results for equation 2, the base model, are shown in the first column. The coefficients on all four of the control variables are statistically significant. Equity grants in the current year are positively correlated with those from the previous year, and with firm size. CEOs receive higher grants than other executives. Grants are negatively related with financial leverage. The signs for these parameters are the same as for the 8

9 models of bonus compensation. The parameter λ 1 is not statistically significant, suggesting no relationship between ROE and equity grants. As expected, the grants are positively correlated with revenue growth. In this model the most curious result is a negative relationship between stock return and equity grants. Recently, Lord and Saito (2012) also find a negative relationship between (proportional) option grants and stock returns for a sample of non-financial firms. Normally, equity grants are considered a reward for firm performance. But, an alternative view is that they are an incentive for better performance in the future. This might suggest that managers of firms with poor stock performance be given more stock-based pay. The empirical results for equation 4 to test the first hypothesis, that executive equity grants are as sensitive to negative firm performance measures as to positive ones, are given in the second column of Table 3. The coefficient estimates on the control variables are similar to those for the base model. As in the base model, the parameter β 1 on stock returns is significantly negative. The coefficient, β 3, on the interactive term is not statistically significant, meaning that the negative relationship between managerial equity grants and share returns holds equally when returns are negative as when they are positive. The estimate for β 2, on the dummy variable, is also negative and significant, suggesting that equity grants are somewhat lower when stock returns are negative. The parameter λ 1 is positive and significant, and λ 3 is negative, significant, and of roughly the same magnitude as λ 1. This implies that managerial equity grants rise with ROE when earnings are positive, but do not fall with earnings when ROE is negative. This is exactly like the relationship between executive bonuses and ROE. The relationship between equity grants and revenue growth is also similar to that for bonuses. The parameter δ 1 is positive and significant, so when revenues are growing, grants increase with firm size. But, δ 3 is negative and of a greater magnitude than δ 1. Because revenue growth is negative in this range, equity grants rise as firm size shrinks. The coefficient δ 2 is significant and positive, suggesting that equity grants are a bit higher when revenues are declining. Results for equation 6 to test the second hypothesis, that performance is the same during and after the crisis as before, are shown in the third column of Table 3. Again, the sign of the coefficients for the control variables are similar to those in the base model. In this case, µ 1 is positive and significant, suggesting, that all else the same, equity grants are actually higher during and after the crisis than before. This is presumably because of the decline in option grants after the dot.com crash near the turn-of-the-century, and the marked increase in restricted stock grants in the last few years. In this specification, the parameter β 1 on stock returns is again negative. The coefficient β 4 is negative, significant and of roughly the same magnitude of β 1. This means that there is no linear relationship between managerial equity grants and stock return during or after the crisis. The coefficient λ 1 is positive and significant, and λ 4 is negative, significant and of roughly the same magnitude as λ 1. This implies that before the crisis there was a positive relationship between equity grants and ROE, but, during the crisis years and after this relationship disappears. The parameter δ 1 is positive and significant, meaning that before the crisis there was a positive relationship between revenue growth and executive equity grants. The coefficient δ 4 is also positive, meaning the positive relationship becomes even stronger during and after the crisis. Because almost 54% of the firms exhibited declining revenues after the beginning of the crisis, this implies that equity grants were probably falling sharply when firm revenues declined. 9

10 TABLE 3 RELATIONSHIP BETWEEN EXECUTIVE EQUITY GRANTS AND FIRM PERFORMANCE MEASURES Equation 2 Equation 4 Equation 6 Coefficient Estimate T-Value Estimate T-Value Estimate T-Value Intercept α 0 3, *** *** 2, *** LEQGR α *** *** *** DCEO α *** *** *** CAP α 3-4, *** -3, *** -3, *** LREV α *** *** *** D_2006 µ *** STKRET β *** *** *** D_STKRET β *** N_STKRET β T_STKRET β *** ROE λ , *** 1, *** D_ROE λ N_ROE λ 3-1, *** T_ROE λ 4-2, *** REVGR δ *** *** *** D_REVGR δ *** N_REVGR δ 3-1, *** T_REVGR δ *** Obs 5,940 5,939 5,941 Adj R F-Value 1,697.46*** *** 1,068.18*** *** Significant at 99% Level, ** Significant at 95% Level, * Significant at 90% Level, EQGR - Annual Executive Equity Grants (Options and Restricted Stocks). LEQGR Lagged Annual Equity Executive Grants. DCEO One/Zero Dummy Variable set to One if the Executive is the CEO. CAP Capital Structure (Ratio of Liabilities to the Market Value of Assets). LREV Logarithm of Revenues. STKRET Annual Stock Return. ROE Return on Equity. REVGR Percentage Growth in Revenues. D_STKRET - One/Zero Dummy Variable set to One if Stock Returns are Negative. N_STKRET - Interaction Between STKRET and D_STKRET. D_ROE - One/Zero Dummy Variable set to One if Return on Equity is Negative. N_ROE - Interaction Between ROE and D_ROE. D_REVGR - One/Zero Dummy Variable set to One if Revenue Growth is Negative. N_REVGR - Interaction Between REVGR and D_ REVGR. D_ One/Zero Dummy Variable set to One if the Observation is in the Fiscal Year 2007, 2008 or T_STKRET - Interaction Between STKRET and D_2007. T_ROE - Interaction Between ROE and D_2007. T_REVGR - Interaction Between REVGR and D_

11 6. Summary My first hypothesis is that the compensation of bank executives is not as sensitive to poor firm performance as to positive results. Both bonuses and equity grants are positively correlated with ROE when earnings are positive, but this relationship disappears when ROE is negative. This is thought to be common of the design of bonus contracts. But, it is interesting that equity grants also seem to be insensitive when earnings are negative. Bonuses are uncorrelated with positive stock returns, but they decline when share prices fall. Bonuses and equity grants have a complex relationship with growth in revenues, but it seems that both forms of performance-based compensation increase when the level of revenues change, be it rise or fall. There is little reward if revenues remain relatively constant. While bonuses and equity grants are both insensitive to declines in earnings, the same does not seem the case for the effects of stock returns and revenue growth on the compensation of bank executives. The second hypothesis is that the design of executive compensation grants has remained unchanged by the crisis. Both bonuses and equity grants are positively related to ROE before the crisis. But, during and after this relationship disappears. Interestingly, before the crisis, equity grants are negatively correlated with stock returns. This may mean that these grants are used primarily as inducements for managers of poorly performing banks to improve. But, during and after the crisis this negative relationship disappears. In contrast, bonuses are positively related with stock returns, both before and after the crisis. Both bonuses and equity grants are positively related with revenue growth before the crisis. However, after the crisis, the relationship of bonuses with firm growth turns negative, and that with equity grants becomes more strongly positive. Therefore, the crisis has clearly had a major impact in how performance-based pay is granted to bank executives. However, as the economy slowly recovers, it remains to see whether these changes will have some lasting effects, or will quickly (and quietly) pass. 11

12 References Bebchuk, L.A. & Fried, J.M. (2004). Pay Without Performance: The Unfulfilled Promise of Executive Compensation. Cambridge, Mass., Harvard University Press. Bebchuk, L.A., Fried, J.M. & Walker, D.I. (2002). Managerial Power and Rent Extraction in the Design of Executive Compensation. University of Chicago Law Review, 69, Black, F. & Scholes, M. (1973). The Pricing of Options and Corporate Liabilities. Journal of Political Economy, 81, Cai, J. & Milbourn, T. (August 30, 2010). Bank Executive Pay. Economic Trends, Federal Reserve Bank of Cleveland: Gabaix, X. and Landier, A. (February 2008). Why has CEO Pay Increased So Much? Quarterly Journal of Economics, 123, Goldman, H. & Moore, M.J. (February 24, 2011). Wall Street Bonuses Decline 8% in 2010, New York Comptroller DiNapoli Says. Bloomberg.com new-york-s-dinapoli-says.html Hall, B.J. & Liebman, J.B. (1998). Are CEO s Really Paid Like Bureaucrats? Quarterly Journal of Economics, 63, Hodgson, P. (November 2010). Wall Street Pay: Size, Structure and Significance for Shareholders. White Paper, Council of Institutional Investors. Holstrom, B.R. (Spring 1979). Moral Hazard and Observability. Bell Journal of Economics, John, K. & Qian, Y. (April 2003). Incentive Features in CEO Compensation in the Banking Industry. Policy Review Federal Reserve Bank of New York, Leone, A.J., Wu, J.S. & Zimmerman, J.L. (2006). Asymmetric Sensitivity of CEO Cash Compensation to Stock Returns. Journal of Accounting and Economics, 42, Lord, R.A, & Saito, Y. (2010). Trends in CEO Compensation and Equity Holdings for S&P 1,500 Firms: Journal of Applied Finance, 20, Lord, R.A, & Saito, Y. (2012). Does Compensation Structure Alleviate Personal CEO Risk? Journal of Business Finance and Accounting, forthcoming. Murphy, K.J. (1999). Executive Compensation. Handbook of Labor Economics. Volume 3B, Edited by Ashenfelter, O. and D. Card. Elsevier Press. Amsterdam: 2,485-2,

13 Obama, B.H. (June 17, 2009). Remarks by the President on 21 st Century Financial Reform. The White House website: President-on-Regulatory-Reform/. Smith, C.W. & Stulz, R.M. (December 1985). The Determinants of Firm Hedging Policies. Journal of Financial and Quantitative Analysis, 20, Smith, C.W. & Watts, R.L. (1992). The Investment Opportunity Set and Corporate Financing, Dividend and Compensation Policies. Journal of Financial Economics, 32, United States Department of Treasury (February 4, 2009). Treasury Announces New Restrictions on Executive Compensation, Press Room, U.S. Department of Treasury website: Welsch, R.E. (1980). Regression Sensitivity Analysis and Bounded-Influence Estimation. Evaluation of Econometric Models. Edited by Kementa, J., and J.B. Ramsey. Academic Press. New York: