CEO Compensation and Bank Mergers

Richard T. Bliss Babson College Babson Park, MA 02157 Richard J. Rosen Kelley School of Business Indiana University Bloomington, IN 47405

This draft: February 2000 First draft: May 1998 Abstract Recent bank mergers generally did not improve relative operating performance or produce positive abnormal returns to acquiring bank shareholders. We examine the relationship between mergers and CEO compensation during 1986-1995, a period marked by overcapacity and frequent mergers. We find that mergers have a net positive effect on compensation, mainly via the effect of size on compensation. Compensation generally increases even if mergers cause the acquiring bank's stock price to decline, as is typical after a merger announcement. The form of compensation affects merger decisions, since CEOs with more stock-based wealth or compensation were less likely to make an acquisition.

JEL classification: G21, G34 Keywords: banking, mergers, compensation

We would like to thank Vladimir Shevtsov and Alex Pardo for research assistance. Useful suggestions were given by the referee, Gordon Hanka, Nathan Stuart, and participants in presentations at the 1999 Western Finance Association Meetings, the Kelley School of Business, and the Federal Reserve Bank of Chicago Bank Structure and Competition conference. Contact author: Rosen at (812) 855-3416, FAX (812) 855-5875, email: [email protected]. Bliss at (781) 2395883, FAX (781) 239-6465, e-mail: [email protected].

CEO Compensation and Bank Mergers 1. Introduction In the late 1980s and the 1990s, the banking industry was in the midst of a rapid consolidation, with a record number of mergers contributing to a one-third reduction in the number of independent banks between 1985 and 1995 (see Figure 1 and Berger, et. al., 1999). 1 This paper explores the relationship between mergers and CEO compensation during this period. We describe, at least in part, the private benefits available to managers from an acquisition and ask how these private benefits affect the decision to engage in a merger. There are many other reasons why bank consolidation may have occurred. During the 1980s and 1990s, there were a number of big changes in the competitive environment of the financial services industry. Banks were facing increasing competition from non-bank competitors (Gorton and Rosen, 1995), a fact evidenced by the decline in the share of nonfinancial short-term corporate debt held by banks (see Figure 1). Regulations against interstate (and, in some cases, intrastate) expansion were repealed (Hubbard and Palia, 1995) and regulations restricting bank underwriting activities were relaxed. The importance of information systems may also have introduced new economies of scale into banking (Radecki. et. al, 1997). These events may have created overcapacity in banking, leading to a situation in which (the right) mergers could improve the overall efficiency and profitability of the industry (Gorton and Rosen, 1997). The empirical evidence on whether the right mergers took place is mixed. There is evidence that, in general, efficient banks took over less efficient banks (Berger and Humphrey, 1992) and that mergers between banks in the same geographic area increased market power for the survivors (Prager and Hannan, 1999). But this is not the same as saying that mergers improve performance. There also is evidence that, on average, the mergers did not improve bank operating performance or produce positive abnormal returns to shareholders (Berger and Humphrey, 1992; Houston and Ryngaert, 1994; Rhoades, 1994). Calomiris (1999) argues that these studies of individual bank mergers can miss overall gains from mergers, especially during a merger wave. Mergers may also be a response to inefficiency (or they may cause other banks to need to improve efficiency). According to Calomiris, studies of individual bank mergers may miss the improvements after a merger because they compare merging banks to non-merging banks. If the non-merging banks are also responding to the same pressures by becoming more efficient through internal means, then the studies will show no gain when, in fact, mergers are

-2 -

the best way for some banks to become more efficient. Consistent with this view, banks have had record levels of profitability in the 1990s. We take no position on this debate; rather, we point out how private benefits can affect and be affected by the decision to make an acquisition. Previous studies identify possible links between mergers and managerial self-interest. These include hubris (Roll, 1986), diversification of personal risk (Amihud and Lev, 1981; Morck, Shleifer and Vishny, 1990; May, 1995), and the additional compensation and perquisites associated with size (Murphy, 1985; Jensen and Murphy, 1990). In this paper, we evaluate the last explanation by analyzing the relationship between mergers and compensation of bank CEOs between 1986 and 1995. There is extensive evidence that CEOs of larger firms earn more (Jensen and Murphy, 1990; Hubbard and Palia, 1995). We contribute to this literature by specifically examining how growth via mergers adds to compensation. Once we have established that mergers add to CEO compensation, we look at the flip side of the issue: whether the form of a CEO’s compensation affects his decision to acquire another firm. Mehran (1995) shows that firm performance is related to the form of the compensation contract. This suggests that CEOs may take compensation into account when taking strategic actions. We test whether this is true for banking mergers. Two factors known to influence managerial compensation are firm size and stock price performance. 2 Merger decisions significantly affect both, so we must consider their effect on CEO compensation when we examine the overall effect of mergers. Previous studies examining the effect of size on compensation have not distinguished among the different ways that firms can grow. Specifically, mergers are a relatively fast way to increase the size of a firm. If Boards of Directors reward managers equally for all types of growth, then a manager desiring a quick increase in compensation will have a strong incentive to make acquisitions. Recognizing this, Boards may choose to reward some forms of growth more than others based on the perceived contribution to shareholder value. We test this by distinguishing merger growth from other growth. We find that size – no matter how acquired – adds to CEO compensation. We also look at the effect of bank mergers on the acquirer’s stock price. We find a consistent negative announcement reaction, a result consistent with prior research. This drop in stock price reduces both the compensation and wealth of the acquiring bank’s managers. However, these losses are offset by the increases in CEO compensation from the effect of the merger on bank size. The net result is that even mergers which reduce shareholder value can be in a manager’s private interest. 3

-3 -

To illustrate how asset growth and stock price changes affect compensation, consider the case of two banks headquartered in North Carolina: First Union and Wachovia. These were two of the three largest North Carolina banks at the start of our sample period (NCNB which became NationsBank was the other). Between 1986 and 1995, First Union's aggressive acquisition strategy resulted in an annual asset growth rate of 17.3%. Over the same period, Wachovia's assets grew at a 9.2% annual rate, with only one large merger. On the other hand, Wachovia's stock price grew at a 17.1% annual rate, significantly greater than First Union's 14.5% annual rate. Thus, First Union grew more in size while Wachovia had the better stock performance. In this case, it appears size was more important in determining compensation as total CEO compensation rose at an annual rate of 13.2% at First Union, but only 11.9% at Wachovia. Acquisitions lead to higher CEO compensation, but that does not imply that merger decisions are made because of their effect on a CEO’s wealth. To test this, we ask whether CEOs with a greater percentage of cash compensation relative to stock-based compensation and more stockbased wealth are more likely to make acquisitions. We find that higher levels of stock-based compensation and wealth reduce the probability that banks make acquisitions. This is consistent with the idea that CEOs respond to incentives and make fewer wealth-reducing mergers when they own more stock. In the next section, we discuss the relevant literature on compensation and mergers. Section 3 describes the data used in our analyses. The empirical results appear in Sections 4 and 5, with Section 4 describing the effect of mergers on compensation and Section 5 the effect of compensation on mergers. Concluding remarks are in Section 6. 2. Literature review and hypotheses This paper connects two strands of literature that have received attention in recent years. The first is research on executive compensation. Work in this area evaluates the agency relationship between managers and owners, and asks how compensation can be used to mitigate potential conflicts. The empirical work on compensation explores the determinants of executive pay, including CEO traits, firm performance, and industry and firm characteristics. The second strand of literature we incorporate into our research considers operating efficiencies and value creation in bank mergers. Specifically, we are interested in whether or not these factors provide motivation for the large number of observed mergers.

-4 -

2.1. Executive compensation Since at least Jensen and Meckling (1976), researchers have examined the relationship between firm ownership and control. When owners cannot observe managerial effort, it is beneficial to link part of executive pay to a proxy for effort, typically some measure of firm performance (Holmstrom, 1979; Grossman and Hart, 1983). This measure can be accounting- or market-based. In cases where accounting data are noisy or can be easily manipulated, or where managerial decisions significantly affect firm value, stock-based compensation will be more effective (Holmstrom, 1979; Lambert and Larcker, 1987). Tying a manager’s compensation to firm performance, usually by making a significant percentage of total pay equity-based, is one way to overcome agency costs and to motivate value-maximizing behavior (Hirshleifer and Suh, 1992). The empirical research finds that the most consistent determinants of executive compensation are stock returns and firm size (see Murphy, 1998, for a review of the compensation literature). Jensen and Murphy (1990) examine the pay-for-performance relationship (the connection between executive compensation and shareholder wealth) for a sample of manufacturing firms between 1969-83. They find that, on average, CEOs receive just $3.25 for every $1,000 increase in shareholder value. This $3.25 is heavily weighted towards market-based components; only a small fraction (approximately $0.02) is current salary. Hall and Liebman (1998) find that pay-performance sensitivity has increased since 1980, due primarily to an increase in stock-based pay. These papers show the impact that stock price changes can have on compensation. Many studies, including most of those cited above, report a strong link between firm size and managerial rewards. 4 This is not surprising, since as Murphy (1998) points out, most large firms set compensation by looking at the compensation of peer group executives, and size is a determinant of which firms are in a peer group. There are several theoretical explanations for the positive correlation between CEO pay and firm size. Compensation can be used to motivate effort among lower-level managers who view the top job as spoils that go to the winner of an intra-firm tournament (Nalebuff and Stiglitz, 1983; Rosen, 1992). A bigger firm represents a larger tournament, and therefore demands a commensurate prize. Also, managing a bigger firm might involve more skill than managing a smaller firm. Here, compensation is used to solve the adverse selection problem in choosing a manager. These explanations for the correlation between compensation and size imply that

-5 -

better managers control bigger firms, not that making a firm larger should increase the compensation of an existing manager. Corporate governance concerns also can explain why a manager could receive a higher salary as his firm grows. Larger firms tend to have more diffuse ownership. This is especially true when a firm grows by using stock to pay for acquisitions (as happens for most of the mergers in our sample). After a stock-based merger, managers may be able to consume more perquisites, including compensation, with lower probability that stockholders will monitor and discipline them. 5 There are several studies that focus on executive compensation in the banking industry. Barro and Barro (1990) use the Jensen and Murphy (1990) methodology to confirm a positive relation between pay and performance using a sample of commercial banks. Crawford, Ezzell, and Miles (1995) and Hubbard and Palia (1995) both consider the effect deregulation of the banking industry had on CEO compensation and turnover. Both find a more sensitive pay-forperformance relationship after deregulation, and Hubbard and Palia (1995) report a significant increase in CEO turnover following deregulation. Both papers note that the observed increase in pay-performance sensitivity is consistent with theoretical research on the principal-agent relationship. Finally, Houston and James (1995) show that while bank CEOs receive a smaller percentage of their total pay from equity-based rewards than CEOs in other industries, compensation policies in banking are more sensitive to firm performance. Cross-sectional differences in managerial ability and effort are primarily reflected in a firm's idiosyncratic return, that is, in its performance relative to some industry benchmark. This suggests that compensation contracts designed to reward superior performance by executives place a greater weight on firm-specific, rather than industry-wide, changes in value. Gibbons and Murphy (1990) examine a sample of industrial firms and find empirical support for this hypothesis. However, Barro and Barro (1990) report that in their sample of banks, relative performance is rewarded at the same level as industry performance while Hall and Liebman (1998) report that relative performance is not a significant component of CEO compensation packages. These mixed results lead us to include both firm-specific and relative performance measures. The empirical studies cited above all focus on the effect of performance and other factors on compensation. An underlying assumption is that it is important to have compensation depend on performance because of the incentive effects of pay on managerial actions. For example, Jensen and Murphy (1990) say that stock-based compensation “provides incentives for CEOs to take

-6 -

appropriate actions.” Mehran (1995) finds that firm performance improves measurably when firm managers have a higher percentage of stock-based compensation. This suggests that the form of the compensation contract might affect important managerial decisions such as acquisitions. 2.2 Mergers and value creation A frequent justification for mergers is that they reduce costs and improve operating efficiency, which in turn increases shareholder returns. 6 As discussed earlier, the empirical support for this claim in the existing research on bank mergers is weak. Berger and Humphrey (1992), Linder and Crane (1993), Rhoades (1994), and Srinivasan and Wall (1992) all find that there are typically no cost efficiencies resulting from bank mergers. When costs fall, it is often because banks shed assets after mergers. Another possibility is that the diversification inherent in mergers is valuable to shareholders. Gorton and Rosen (1997) show that during periods of industry consolidation, diversification is advantageous. Mergers may produce wealth gains even without cost efficiencies by increasing diversification. Akhavein, Berger, and Humphrey (1997) and Demsetz and Strahan (1997) find that bank mergers serve to diversify banks, thereby allowing them to take on more investment risk for a given level of firm risk. However, there is no evidence of a link between this incremental diversification and increased shareholder returns. Even if mergers create value through efficiency gains, cost savings, or diversification benefits, the acquiring bank’s shareholders only benefit if the price paid for the target is sufficiently low. If all of the merger benefits are passed on to the shareholders of the target firm through a high acquisition price, then shareholders in the acquiring firm will not want their CEO to make the acquisition. Research on stock price changes around an announcement suggests that acquisitions lead to a decline in acquirer stock prices. For example, Hannan and Wolken (1989) show that a merger announcement reduces the stock price of the acquirer by 3.8% while Houston and Ryngaert (1994) find a 2.8% reduction. Note that the net effect of a merger on total acquirer and target value is not necessarily negative since target firm stock prices generally rise (by 11.1% in Hannan and Wolken's sample and 10.3% in Houston and Ryngaert’s). 7, 8 Given these interrelationships among pay, performance, and size, it is natural to consider the effect of mergers on executive compensation. In this paper, we look at the impact of mergers on compensation at the acquiring firm, that is, for those managers responsible for initiating the merger. Does the structure of compensation contracts motivate a manager’s decision to engage

-7 -

in a potentially value-reducing merger as suggested in Morck, Shleifer, and Vishny (1990)? After all, mergers can provide a quick way for managers to increase bank size, and therefore, possibly their pay. We need to look at the marginal contribution of mergers, since Boards of Directors could structure compensation contracts to reflect expected acquisition policies. Avery, et. al. (1998), using a large sample of industrial firms, report that the average level of pay is the same at firms that merge and those that do not. Still, managers at the acquiring firms could be rewarded on the margin for an acquisition. Finally, we examine whether compensation influences merger decisions, since the cost to shareholders of CEO compensation is small relative to the potential losses from a bad acquisition decision. 3. Data The data used in our analyses include information on bank size, mergers, managerial compensation, and stock price performance. We consider the ten-year period from 1986 to 1995, a decade marked by significant consolidation in the banking industry. The period also includes a full economic cycle for banks. Our sample consists of all bank holding companies that were among the 30 largest in asset size in at least one year of the sample period and existed for at least five years during the sample period. 9 This results in a sample of 32 banks and 298 bank-years. Data on managerial compensation is collected from proxy statements and includes information on salary and bonus, beneficial share ownership, restricted shares, and stock options granted and owned. Proxy statements are also the source of information on insider shareholdings. Information on bank merger activity is compiled from a number of sources, including Securities Data Corporation, Lexis-Nexis, Mergers and Acquisitions magazine, and the Federal Reserve Board. Control variables for the size and growth rate of bank assets are from Compustat. Stock price data come from the CRSP database. Panel A of Table 1 lists the banks included in the final sample, and Panel B provides descriptive statistics. Mean and median asset size for the sample banks are $64.4 billion and $44.0 billion respectively (all values are reported in 1995 dollars). The mean stock market equity for the banks is $4.7 billion with a median value of $3.3 billion, with similar values for accounting equity. Top managers are paid in many different ways. For some CEOs, salary forms only a small part of their total compensation. Managers may also receive performance-related bonuses, restricted shares, and stock options. We want to test the relationships between managerial compensation, firm size, merger activity, and shareholder returns. We also want to distinguish

-8 -

between the fixed and variable (i.e., performance-based) components of managerial pay, since the incentives for a manager to increase firm value depend on whether or not pay is related to such value creation. Following others such as Hubbard and Palia (1995), we break compensation into two parts. Our first measure of compensation, cash payments, includes the CEO’s salary, bonus, and any other cash payments as reported in the bank’s proxy statement. The second measure is total compensation, which is cash payments plus the value of newlygranted restricted shares and stock options. The latter includes the value of stock options and restricted shares granted during the year. Options are valued using the Black-Scholes formula assuming a 10-year maturity for all options. 10 Stock price volatility is estimated from daily stock returns in the year of the grant. Note that we do not include changes in the value of existing shares and stock options in total compensation, only the value of new equity-based compensation. Panel B of Table 1 shows that cash compensation averages $1.61 million, which represents 66 percent of total new compensation. Clearly, restricted shares and stock options are important factors when evaluating managerial incentives to increase compensation. Compensation varies widely across banks and time, which is not surprising since most banks grew significantly over the sample period. Even after normalizing compensation for asset size, there is still a wide variation. Total compensation ranges from $6 to $206 per million dollars of assets, while cash compensation ranges from $4 to $139 per million dollars of assets. Managerial decisions may depend not only on new compensation, but also on other stockbased holdings. We have information on the shares and options controlled by a CEO. This is not sufficient to provide a measure of CEO stock-based wealth, since we do not have the strike prices for all options. However, we can estimate the marginal effect of a change in the stock price by assuming a fixed marginal change in wealth with a change in stock prices for all the options. Following others (e.g., Gilson and Vetsuypens, 1993), we assume that options increase in value by 60 percent of the change in the stock price (Murphy, 1998, shows that this approximates the change for an option that is just in the money). 11 All of our qualitative results are robust to increases in the option sensitivity up to parity with stock. We find that, on average, CEOs gain $104,608 in wealth for every one percent increase in stock price. The sensitivity of wealth to stock price changes varies greatly across the sample, from a minimum of $2,688 to a maximum of over $750,000 per percent change in price. Table 2 shows data on the acquisitions by the sample banks during 1984 - 1995. We include 1984 and 1985 since we look at the impact of the prior three years of mergers and growth on

-9 -

compensation. We focus on large mergers, that is, those most likely to impact CEO compensation and therefore influence the decision to acquire. We call these “megamergers,” and define them as acquisitions where the total assets of the target bank immediately before the merger announcement are at least ten percent of the acquirer’s pre-merger assets and where the target is not a failed bank. 12 There are 66 megamergers over the 12 years. Table 2 also includes data on all mergers, and to ensure robustness, we test all hypotheses using both samples. Table 2 shows that more than 75% of the total assets acquired were through megamergers. For these transactions, the average target bank has assets of $11.9 billion (in 1995 dollars), which is slightly less than one-half the size of the mean acquirer. Acquisitions accounted for 3.6 percentage points of asset growth per year, which is over one-half of the total growth of the sample banks. Note that the average stock price reaction to a merger announcement is -2.4%, a result consistent with other studies. 4. The effect of mergers on compensation If asset growth via merger increases compensation, then all else equal the compensation at high-merger banks should grow more rapidly than at low-mergers banks. To test this, we divide our sample into two groups based on the dollar value of total merger activity during the sample period. The 16 most active acquirers are in the “high-merger” group and the 16 least active acquirers are in the “low-merger” group (Panel A of Table 3 gives the banks in each group). As Panel B of Table 3 shows, high-merger banks have significantly more merger activity than the low-merger banks even though they are slightly smaller. While the level of total compensation is similar across the groups, the CEOs of high-merger banks receive a significantly larger proportion of their compensation in cash, a result analyzed further below. The other characteristics of the banks in the different groups, however, are similar. The first row of Table 4 presents the average increase in CEO compensation over the sample period. In calculating this, we exclude a CEO’s first year (that is, the gain when he is promoted to CEO). As the first row of Table 4 shows, the mean gain in total compensation was 10.3% per year and 141.6% over the entire sample period. Comparable figures for cash compensation are 9.9% per year and 133.9% over the entire sample period. Comparing the high- and low-merger groups, as shown in the next two rows of the table, we see that total compensation rises slightly more and cash compensation rises significantly more at high-merger banks than at low-merger banks. This provides some evidence as to the importance of mergers to compensation. 13

- 10 -

To isolate the role of merger growth, we need to control for other factors that could influence compensation. Our basic empirical model to test the relationship between mergers and compensation is: Compensation = f (merger activity, firm performance, size, control variables)

(1)

We use two proxies for merger activity. The first, MEGA3, is the assets acquired through megamergers in the three years prior to the end of year t. To control for potential heteroskedasticy in the sample, we normalize this and all other dollar values by dividing by the total assets of the bank. The second proxy, MRG3, represents the assets acquired through all mergers in the three years prior to the end of year t. In computing MEGA3 and MRG3 for year t, we assume that the assets acquired in years t-1 and t-2 grow at the same rate as the bank’s other (non-acquired) assets. The variables MEGA3 and MRG3 allows us to estimate how much acquiring a dollar of assets via merger in the any of the previous three years impacts compensation in the current year. We use a three-year window because it may take time for a CEO’s compensation to rise to its steady-state level after an acquisition. Following others, we measure firm performance using both stock market and accounting data. We use two variables to proxy for stock-market equity value changes, primarily to distinguish between industry-wide and firm-specific movements. INDEX CHG measures the change in the bank’s equity value relative to changes in an index of bank stocks. It has mean value $665 million and a median value of $292 million per percentage point change in the index. The mean and median percentage increases are 18.3% and 9.6%. The bank stock index is an equally-weighted index of the top 100 banks, but our results are robust to using other indices. FIRM CHG is the change in the bank’s equity value not due to changes in the index value. It is defined as the total change in value minus the change in value due to movement in the index. We also want to have a control for accounting return. We use return on assets (ROA) since firm compensation policies often explicitly tie compensation to ROA (Murphy, 1998). Banks in the sample have a mean and median ROA of 0.87% and 1.04%. Since previous work has found that pay is related to firm size, we include variables for bank size and a proxy for non-merger growth. RTA3 in year t is defined as the bank’s real total assets three years prior to the observation year, that is, at the start of year t-2. GROW3 is the dollar growth of the assets that were in place at the start of year t-2 plus the growth in acquired assets. Note that RTA3 + GROW3 + MEGA3 sums to the bank’s total assets in year t. For the average sample bank, the mean asset value of $64.4 billion shown in Panel B of Table 1 includes $4.3

- 11 -

billion of asset growth and $6.2 billion of assets acquired through megamergers in the previous three years. The comparable median values are $44.0 billion of total assets, $2.4 billion of growth, and $0 of acquired assets (indicating that in any three-year period, fewer than half of the sample banks made acquisitions). 14 The mean and median growth rates are 4.2% and 2.3% for asset growth and 6.0% and 0% for megamerger growth. Other regression variables are meant to control for differences across firms in compensation policies and managerial entrenchment. These include CEO age (AGE, with mean and median of 57 years old) and CEO share ownership as a percentage of total shares outstanding (CEO SHR, with a mean of 0.303% and median of 0.250%). We also looked at other measures of CEO control, such as the years as CEO, and variables that measure how close a CEO was to retirement. None of these are used in the final analyses since they added little, if any, explanatory power. We control for macroeconomic effects and regulatory changes using year dummies, which are not reported in the tables. The idea is that Boards of Directors may not hold poor performance against a CEO if the performance was due to factors beyond the CEO's control. We also looked at geographic diversification. The private benefits of a CEO might be increasing (Amihud and Lev, 1981) or decreasing (Gorton and Rosen, 1997) in the diversification of a bank. Rose and Shepard (1997) show that there is a diversification premium in CEO compensation, although they attribute this more to needing to attract better managers for more diversified (and hence, more complicated to manage) firms. We tested for the impact of diversification in our sample by looking at a measure of geographic diversification and how it changes upon an acquisition (since almost all of our acquisitions provided little product diversification). The results were never significant, and so they are not reported. This is similar to Houston and Ryngaert (1994) who find no difference between the stock price reaction to interstate and intrastate bank mergers. 4.1. Full-sample results Table 5 presents the full-sample regression of (1) using total new compensation (columns 12) and cash compensation (columns 3-4) as dependent variables. The broad implications of the regressions support previous findings that both asset size and stock price performance matter in determining compensation. Asset growth increases both cash and total compensation. For every $1 million increase in assets, a CEO gets $24-59 more in total compensation including $6-32 of cash compensation. Moreover, mergers increase compensation faster than internal growth. For

- 12 -

example, a bank that acquires $1 million in new assets through a megamerger will increase the total compensation of its CEO by $54 while $1 million of internal growth increases CEO total compensation by $30. This difference is statistically significant at the 5% level. The regressions indicate that it is not just that CEOs of larger banks are paid more; it is also true that making a bank larger generally increases pay. After controlling for size, a $1 million increase in the equity value of a bank from an industry-wide change in stock prices increases CEO total compensation by $952 while a $1 million increase from an idiosyncratic change increases total compensation by $536 (this uses the values from column 1, the megamerger regression). With similar increases, cash compensation rises by $284 and $156, respectively (column 3). These results are consistent with the findings of Barro and Barro (1990) and similar to those in Hubbard and Palia (1995). The regression results can be used to predict the size of the actual changes in CEO compensation for the sample megamergers. For each merger in the sample, we estimate the change in compensation using two assumptions on the impact of the merger on equity value. First, we assume that the change in equity value is given by the cumulative abnormal return (CAR) from the day prior to the announcement to the day following the announcement. If we are looking for whether private benefits motivate mergers, this is equivalent to assuming that the CEO knows how the market will react to the announcement. We also estimate the change in compensation using the mean CAR for the sample. If a CEO does not know how the market will react to his bank’s merger, he may project the effect on his compensation using the mean CAR. The estimated changes in compensation are given in Panel A of Table 6. The median change in compensation from a merger is 20-30% of a CEO’s premerger compensation. In fact, there was only one merger which led to a predicted decline in compensation, and that was only if we use the actual CAR. The regression results suggest that over three-quarters of the mergers in the sample led to at least a 10% increase in compensation. Thus, we can see that mergers, or at least the large megamergers, result in economically significant increases in compensation for most CEOs. A CEO almost always can expect a large increase in his compensation even when the stock market does not like the merger and the bank’s shareholders experience a loss. This analysis leaves out one important factor: even though the CEO gains from a merger by getting more compensation, like the other shareholders he suffers a loss in value on his existing stock in a typical merger. Panel B of Table 6 estimates the first-year change in wealth of the CEO. Again, we use both the realized and average CARs. Using the average CAR, the median decline in wealth is $155,728, about 40% of the median compensation increase. Overall, with

- 13 -

the wealth effect, first-year compensation rises by an average of about 30%, with a median increase of 18%. Over three-quarters of CEOs experience a gain in wealth. When evaluating this, remember that the compensation increase is paid to the CEO annually, while the stock price decline is a one-time loss of wealth. Thus, over a period of time, it appears that acquiring CEOs gain significantly from mergers. For example, with an average age of 57 and assuming a retirement of 65, the present value of the gain in compensation for the median CAR is close to $2 million over a CEO’s remaining career. This is much larger than the one-time loss in wealth of $155,728 that results from the merger announcement. We now turn to the other independent variables in Table 5. Since cash compensation often explicitly depends on ROA, it is not surprising that changes in ROA can be important. A one standard deviation increase in ROA (0.67%) increases total and cash compensation by about seven percent each. Compensation is unaffected by CEO age but is higher when a CEO owns a larger share of his bank. An increase in CEO ownership will have approximately a proportionate increase in compensation. So, a one-standard deviation increase in CEO share ownership, 0.20%, increases both total and cash compensation by about 0.20%. Thus, the overall effect on compensation of these variables is small. Finally, the results in Tables 5 and 6 (and the other results in the paper) are robust. We have checked that they are not being driven by a single bank. They are also robust to changes in the functional form of our regression equation. Previous work on executive compensation has generally used one of two functional forms to estimate the relationship between compensation and firm size (Murphy, 1998, has a discussion of the different approaches). The first approach is to regress the dollar value of compensation on the dollar value of firm size, as we do here. The second approach is to regress the log of compensation on the log of size. Murphy says that there is no theoretical reason to prefer one approach to the other. One drawback of using logs is that it is not possible to separate the effects of the components of size, as we want. The most we can test is whether assets acquired through mergers impact compensation more or less than other assets. Using the log-linear approach, however, gives qualitatively similar results, with assets acquired through mergers adding at least as much in compensation as other size. 4.2. High- versus low-merger banks In our sample, some banks make frequent acquisitions, while other banks make few or no acquisitions. If the Board of Directors sets overall merger policy for a bank, then it seems reasonable that the Compensation Committee of the Board will want to reward managers for

- 14 -

hitting acquisition targets. This suggests that the rewards for mergers might differ across banks based on the Boards’ expectations of merger level. To test this, we start with the high- and lowmerger groups from earlier. We want to see if the Boards reward mergers differently in these two groups so we repeat the basic regression in equation (1) for the two groups of banks. The results are reported in Table 7. The results show that the high-merger CEOs are rewarded significantly less per dollar of merger growth than their low-merger counterparts. For $1 million of asset growth from mergers, low-merger CEOs get $40 more in cash and $111 more in total compensation for adding $1 million in assets by merger. High-merger CEOs get much less per dollar of acquisition. These coefficients are insignificant, possibly due to the small sample size. CEOs in the low-merger group receive compensation more closely linked to the bank’s stock price but less tied to accounting return on assets (ROA). For each $1 million increase in equity value, low-merger bank CEOs get between $1025 and $1535 in additional total compensation, while high-merger CEOs get between $292 and $753. Table 7 also shows a significantly larger coefficient on ROA for the high-merger banks. For one standard deviation increase in ROA (0.67%), total compensation rises an insignificant 0.5% for a low-merger CEO versus a significant 11% increase for a high-merger CEO. These split-sample results suggest that the discrepancy in CEO compensation between highand low-merger banks reported in the full-sample regressions may be deceptive. Low-merger CEOs undertake fewer acquisitions, but they are paid more per dollar of assets acquired. They also have compensation that is more sensitive to stock prices. To see the aggregate effect of merger strategy on compensation, we estimate the actual changes in compensation and wealth as in Table 6 for the split sample. The results are reported in Table 8. We report the average change per year compensation (including years for which there was no acquisition), rather than the change per acquisition as in Table 6. This allows us to compare the overall returns from an acquisition strategy. Note that there are a large number of zeros in Panel B because seven of the banks made no acquisitions in the sample period (Table 3, Panel A lists these banks). The aggregate returns to an acquisition strategy vary across the banks. As shown in Table 8, high-merger CEOs gain more compensation per year than low-merger CEOs from their strategy, even if they gain less per dollar acquired. However, once wealth effects are included, the picture becomes less clear. The mean and median compensation plus wealth change is larger at highmerger banks than at low-merger banks if we use actual CARs to estimate the stock price reaction to an acquisition, but lower if we use the average CAR.

- 15 -

4.3 Discussion Acquiring a bank results in an increase in CEO compensation. This leads to two questions: does this potential increase motivate acquisitions decisions and, if so, is this in any way optimal (or second best)? The banking industry during our sample period was facing new competition and had overcapacity. In a consolidating industry such as banking, CEOs have an incentive to acquire other banks rather than have their bank acquired (Gorton, et. al., 1999). On the other hand, mergers can be an efficient response to overcapacity. Boards of Directors want to fashion compensation plans that offer incentives for efficient mergers but do not push CEOs toward bad mergers. Boards also want to ensure that CEOs do not overpay for targets. Our results are consistent with all this. Mergers definitely bring private benefits to CEOs, but the differences between compensation plans at high- and low-merger banks could be a sign that Boards of Directors recognize that not all size is created equal. For CEOs who tend to merge a lot, perhaps too much, the Board shifts the reward from size to profitability, specifically ROA. For other CEOs, such as those that may merge less than the optimal amount, the Board rewards them more for getting bigger by acquisition. This could be a sign that Boards target an optimal level of acquisitions, and structure compensation accordingly. 5. The effect of compensation on mergers In this section, we ask whether the form of CEO compensation and the level of stock ownership affect acquisition decisions. Since mergers typically result in a decrease in the acquirer’s stock price, it follows that CEOs with more equity or whose pay is more heavily weighted towards equity-based incentives might be less likely to make acquisitions. Mehran (1995) finds that firm performance, as measured by Tobin's Q, is increasing in the fraction of the CEO's compensation that is equity-based and in the fraction of the firm owned by the CEO. We loosely follow Mehran when we examine the impact of the form of compensation on merger decisions. We expand on his paper, however, by also looking at the effect of the levels of stock compensation and ownership on these decisions. To illustrate the potential impact of the form of compensation, we ask whether banks that merge frequently have CEOs with higher levels of cash compensation relative to CEOs at lowmerger banks. We divide the sample into high- and low-merger banks as above and examine differences in the average ratio of cash compensation to total compensation. We find that the high-merger bank CEOs have an average of 69% of their compensation in cash versus 62% for the low-merger bank CEOs (see Panel B of Table 3). This 7% difference is statistically

- 16 -

significant. This says that CEOs that acquire more have less of their pay in stock. This is consistent with equity-based compensation serving as a disincentive to acquire. We examine this further by considering the probability of making an acquisition as a function of the form CEO compensation and wealth. In doing this, we use the compensation data for the year prior to a merger announcement. This allows for the usual delay between the announcement and completion of the merger. For some of the mergers, this delay is six months or longer. Bank mergers are typically subjected to additional scrutiny when compared to other mergers because both bank and antitrust regulators must approve every acquisition. Since it is the compensation plan in place at the time of the decision to merge that is important, we use the announcement date rather than the merger completion date. To test the effect of the form of compensation on acquisition decisions we use: Acquisition dummy = f (compensation, wealth proxies, control variables)

(2)

The acquisition dummy is one if a merger announcement is made during year t + 1, and zero otherwise. The independent variables are as of year t, unless otherwise stated. The most important control variable is the mean level of merger activity at a bank over the sample period. The results in Section 4 indicate that Boards of Directors may set compensation policy to affect merger activity. If that is the case, then any correlation between compensation and acquisitions may reflect decisions by the Boards not the incentive effect on CEOs. By including the mean level of merger activity, we can examine how decisions change when the form of compensation changes. Since banks may need time between acquisitions to fully integrate their targets, we also control for recent merger activity. We measure recent merger activity by the percentage of assets acquired through mergers in the past three years. The coefficient on this should be positive if mergers are part of an acquisition strategy and negative if it takes time to integrate targets. We also control for bank performance since banks that are doing better may be more likely to make an acquisition, whether as a reflection of hubris or just because they have more retained earnings (or higher-priced stock). Our performance controls are the percentage gain in stock price over the past year above that on the bank stock index and the market-to-book ratio. We also tried other measures of performance such as ROA, but they were insignificant. Other control variables include a dummy for whether the CEO is at least 50 years old and the fraction of the board that is insiders (management or family of management). These have been found to be important in other studies of corporate control.

- 17 -

Table 9 presents the results of the logistic regressions. We find that the more cash compensation a CEO has relative to total compensation (column 1), the more likely the CEO is to engage in an acquisition. That is, the more stock prices matter to the CEO, the less likely he is to make an acquisition. If mergers generally are wealth-decreasing events for banks, then this is consistent with Mehran's (1995) finding that stock performance is better at firms where managers have a higher ratio of equity-based pay. Mehran looks at compensation ratios only, but it is also possible that the levels of equity ownership and stock-based compensation motivate CEO decisions. To test whether it is levels rather than ratios that matter, we add variables reflecting the dollar value of equity ownership and stock compensation. These are presented in columns 2-4 of Table 9. These results show that stock ownership and compensation reduce the probability that a CEO will engage in an acquisition. These results allow us to narrow down the possible reasons for acquisitions. If diversification of personal risk were important, then CEOs with more exposure to stock prices would be more likely to make acquisitions (Demsetz and Strahan, 1997, show that size and diversification are related, implying that acquisitions generally reduce risk). That is not what we find. While hubris probably plays a role in many corporate decisions, our results suggest that it is not the only motivation for mergers. If it were, then the form of the compensation contract should be unimportant and CEOs with more hubris should make repeated acquisitions (leading to a positive coefficient on previous merger activity). Neither of these is true. The evidence that acquisitions are for job retention is mixed. If job retention dominates other wealth considerations, then the form of the compensation contract and a CEO’s stock ownership should have no effect on the decision to acquire. Again, this is not what we find. On the other hand, if the effect of mergers is to deter acquisition by other banks, then a CEO may not have to make a large number of acquisitions to accomplish this. We find that the probability of an acquisition is decreasing in the dollar value of recent acquisitions, consistent with this. Thus, there is some evidence that the desire to keep their jobs might give some CEOs incentives to make acquisitions. The results shown in Table 9 provide more evidence that the compensation plans set by Boards of Directors have an influence on acquisition decisions. As we saw earlier, Boards set compensation to reduce the rewards from acquisition for high-merger CEOs relative to lowmerger CEOs. Here we see that not just the level of compensation, but also the form of compensation matters. CEOs getting a higher proportion of their compensation as cash or with a

- 18 -

less stock make more acquisitions, leading to a more rapid increase in compensation. This explains the result in Table 4 showing the CEOs of high-merger banks have more rapid compensation increases. Finally, we should be a little cautious before interpreting the negative announcement effects for the banks in our sample as indicating inefficient mergers. The logistic regressions in Table 9 also show that banks may be more likely to make acquisitions when their stock price has risen faster than that of other banks, since the coefficient on the growth above the bank stock index is positive and sometimes significant. This suggests that the adverse selection problem identified by Myers and Majluf (1984), among others, may play a role in acquisition decisions: Bank CEOs are more likely to make an acquisition if they can pay with overpriced stock. Thus, part of the negative announcement effect could be the signal by the acquiring bank that its managers think that their stock is overvalued. An alternative explanation for the negative announcement effects has to do with overcapacity. The market may have expected mergers to occur, although without knowing which firms would be targets and which would be acquirers. The expectation would have been reflected in stock prices. Even if these mergers were efficient ways of reducing capacity, when one bank became an acquirer, its stock price could fall either because the market no longer thought it was as likely to be a target (getting the target premium) or because it overpaid. Nevertheless, we focus on stock-price-decreasing mergers by repeating our analysis after dropping mergers for which there was a positive stock price reaction in the three days centered on the announcement day (not shown). There were nine mergers with positive announcement effects, and excluding them did not qualitatively affect the results presented in Table 9. 6. Conclusion In this paper, we look at the effect of bank mergers on executive compensation during a period of rapid industry consolidation. We find that, on average, acquisitions significantly increase CEO compensation even after accounting for a typical announcement date stock price decline. While the decline in existing wealth partially offsets some of the subsequent salary gains, the vast majority mergers still increases the overall wealth of the CEO, often at the expense of shareholders. There is a documented link between firm size and CEO compensation, and mergers clearly increase firm size. But acquisitions only represent one way that a firm may grow, so we analyze the relationship between CEO compensation and three components of bank size: initial size,

- 19 -

merger growth, and non-merger growth. We find that growth through any means adds similar amounts to compensation, that is, any growth is good for CEO compensation. Thus, one can view mergers as an easy way to rapidly increase compensation. Again, these findings may indicate incentives to merge that ignore the impact of the transaction on the firm’s shareholders. There is some evidence that Boards of Directors partially weigh acquisition patterns when adjusting compensation. CEOs of banks that merge more often are rewarded less per dollar of acquired assets. On the other hand, Boards do not appear to use CEO stock ownership as a tool to affect merger policy. CEOs at high-merger banks a smaller share of equity-based compensation than CEOs at other banks. This may allow these managers to be less concerned than other managers about a negative stock price reaction to their acquisitions. Most previous empirical studies in this area have focused on how size and performance affect pay. We add to this by examining whether compensation affects decisions that may impact performance. We find that CEOs seem to be motivated by the form of their compensation contracts. Banks where CEOs receive more stock compensation or own more stock are less likely to acquire other banks. Since most acquisitions have negative announcement effects, this result is consistent with the theoretical models that show that managers can be motivated by compensation contracts. Finally, while our results indicate that mergers increase compensation and that increasing the stock-based wealth of a CEO reduces the probability that a bank will make an acquisition, they should not be interpreted as saying that private benefits are the cause of bad merger decisions. Banking during the period we studied was being buffeted by the winds of change. These changes caused the industry to have to restructure. The restructuring, of which mergers were a part, may have contributed to the strength of banks in the 1990s.

References Agarwal, Naresh C., 1981, “Determinants of Executive Compensation”, Industrial Relations 20, No. 1, 36-58. Akhavein, Jalal, Allen Berger, and David Humphrey, 1997, "The Effects of Megamergers on Efficiency and Prices: Evidence from a Bank Profit Function," Review of Industrial Organization 12, 95-138. Amihud, Yakov and Baruch Lev, 1981, "Risk Reduction As A Managerial Motive For Conglomerate Mergers," Bell Journal of Economics 12, 605-617.

- 20 -

Avery, Chevalier, and Schaefer, 1998, "Why Do Managers Undertake Acquisitions? An Analysis of Internal and External Rewards for Acquisitiveness", Journal of Law, Economics, and Organization 14. Baradwaj, Babu G., Donald R. Fraser, and Eugene P.H. Furtado, 1990, “Hostile Bank Takeover Offers: Analysis and Implications,” Journal of Banking and Finance 14, 1229-42. Barro, Jason R. and Robert J. Barro, 1990, “Pay, Performance and Turnover of Bank CEOs”, Journal of Labor Economics 8, 448-481. Berger, Allen N., Rebecca S. Demsetz, and Philip E. Strahan, 1999, “The Consolidation of the Financial Services Industry: Causes, Consequences, and Implications for the Future,” Journal of Banking and Finance 23, 135-194. Berger, Allen N., and David B. Humphrey, 1992, “Megamergers in Banking and the Use of Cost Efficiency as an Antitrust Defense,” Antitrust Bulletin 37, 541-600. Calomiris, Charles W., 1999, “Gauging the Efficiency of Bank Consolidation During a Merger Wave,” Journal of Banking and Finance 23, 615-621. Ciscel, David H. and Thomas M. Carroll, 1980, "The Determinants Of Executive Salaries: An Econometric Survey," Review of Economics and Statistics 62, 7-13. Crawford, Anthony J., John R. Ezzell, and James A. Miles, 1995, “Bank CEO Pay-Performance Relations and the Effects of Deregulation”, Journal of Business 68, 231-256. de Cossio, Francisco, Jack W. Trifts, and Kevin P. Scanlon, 1988, “Bank Equity Returns: The Difference between Intrastate and Interstate Bank Mergers”, University of South Carolina Working Paper. Demsetz, Rebecca S. and Philip E. Strahan, 1997, “Diversification, Size, and Risk at Bank Holding Companies”, Journal of Money, Credit, and Banking 29, 300-313. Flannery, Mark J. and Christopher M. James, 1984, “The Effect of Interest Rate Changes on the Common Stock Returns of Financial Institutions”, The Journal of Finance 39, 1141-1154. Gibbons Robert and Kevin J. Murphy, 1990, “Relative Performance Evaluation for Chief Executive Officers”, Industrial and Labor Relations Review 43, 30-51. Gilson, Stuart C. and Michael R. Vetsuypens, “CEO Compensation in Financially Distressed Firms: An Empirical Analysis,” The Journal of Finance 48, 425-458. Gorton, Gary, Matthias Kahl, and Richard Rosen, 1999, "Eat-or-be-eaten: A Theory of Defensive Merger Waves", working paper. Gorton, Gary and Richard Rosen, 1997, "Strategic Mergers for Survival in Banking", working paper. Gorton, Gary and Richard Rosen, 1995, “Corporate Control, Diversification, and the Decline of Banking,” Journal of Finance 50, 1377-1420. Grossman, Sanford J. and Oliver D. Hart, 1983, “Implicit Contracts Under Asymmetric Information”, Quarterly Journal of Economics 98, 123-157. Hall, Brian J., and Jeffrey B. Liebman, 1998, “Are CEOs Really Paid Like Bureaucrats?”, Quarterly Journal of Economics, 103, 653-691. Hannan, Timothy H., and John D. Wolken, 1989, “Returns to Bidders and Targets in the Acquisition Process: Evidence from the Banking Industry”, Journal of Financial Services Research 3, 5-16. Hawawini, Gabriel A., and Itzhak Swary, 1990, Mergers and Acquisitions in the U.S. Banking Industry: Evidence form the Capital Markets (North Holland: Amsterdam). Hirshleifer, David and Suh, 1992, "Risk, Managerial Effort, and Project Choice," Journal of Financial Intermediation 2, 308-345.

- 21 -

Holderness, Clifford G. and Dennis P. Sheehan, 1988, “The Role of Majority Shareholders in Publicly Held Corporations: An Exploratory Analysis; Journal of Financial Economics 20, 317-47. Holmstrom, Bengt, 1979, “Moral Hazard and Observability”, The Bell Journal of Economics, 10, 74-91. Houston, Joel and Christopher James, 1995, “CEO Compensation and Bank Risk - Is Compensation in Banking Structured to Promote Risk Taking?”, Journal of Monetary Economics 36, 405-32. Houston, Joel and Michael Ryngaert, 1994, “The Overall Gains From Large Bank Mergers,” Journal of Banking and Finance 18, 1155-1176. Hubbard, R. Glenn and Darius Palia, 1995, “Executive Pay and Performance: Evidence from the U.S. Banking Industry”, Journal of Financial Economics 39, 105-30. Jensen, Michael C. and William H. Meckling, 1976, "Theory Of The Firm: Managerial Behavior, Agency Costs And Ownership Structure", Journal of Financial Economics 3, 305-360. Jensen, Michael C. and Kevin J. Murphy, 1990, "Performance Pay And Top-Management Incentives," Journal of Political Economy 98, 225-264. Kahn, Charles and Andrew Winton, 1998, "Ownership Structure, Speculation, and Shareholder Intervention," Journal of Finance 53, 99-129. Lambert, Richard A. and David F. Larcker, 1987, “Executive Compensation Effects of Large Corporate Acquisitions”, Journal of Accounting and Public Policy 6, 231-44. Linder, Jane C., and Dwight B. Crane, 1993, “Bank Mergers: Integration and Profitability”, Journal of Financial Services Research 7, 35-55. May, Don, 1995, “Do Managerial Motives Influence Firm Risk Reduction Strategies,” The Journal of Finance 50, 1291-1308. Mehran, Hamid, 1995, “Executive Compensation Structure, Ownership, and Firm Performance”, Journal of Financial Economics 38, 163-184. Morck, Randall, Andrei Shleifer, and Robert W. Vishny, 1990, “Do Managerial Objectives Drive Bad Acquisitions?” The Journal of Finance 45, 31-49. Murphy, Kevin, 1985, “Corporate Performance and Managerial Remuneration”, Journal of Accounting and Economics 7, 11-42. Murphy, Kevin, 1986, “Incentives, Learning, and Compensation: A Theoretical and Empirical Investigation of Managerial Labor Contracts”, Rand Journal of Economics 17, 59-76. Murphy, Kevin, 1998, “Executive Compensation,” working paper, April 1998. Myers, Stewart and Majluf, 1984, “Corporate Financing and Investment Decisions When Firms Have Information That Investors Do Not Have,” Journal of Financial Economics 13, 187-221. Nalebuff, Barry J. and Joseph Stiglitz, 1983, “Prizes and Incentives: Towards a General Theory of Compensation and Competition”, The Rand Journal of Economics 14, 21-44. Prager, Robin A. and Timothy H. Hannan, 1999, “Do Substantial Horizontal Mergers Generate Significant Price Effects? Evidence From the Banking Industry,” Journal of Industrial Economics 46, 433-52. Radecki, Lawrence J., John Wennigner, and D. K. Orlow, 1997, “Industry Structure: Electronic Delivery’s Potential Effects on Retail Banking,” Journal of Retail Banking Serivces 19, 57-63. Rhoades, Stephen A., 1994, “A Summary of Merger Performance Studies in Banking, 1980-93”, Federal Reserve Board of Governors, Staff Study Number 167, July. Roberts, David R., 1956, “A General Theory of Executive Compensation Based on Statistically Tested Propositions,” Quarterly Journal of Economics 70, 270-294.

- 22 -

Roll, Richard, 1986, "The Hubris Hypothesis Of Corporate Takeovers," Journal of Business 59, 197-216. Rose, Nancy L. and Andrea Shepard, 1997, “Firm Diversification and CEO Compensation: Managerial Ability or Executive Effort,” Rand Journal of Economics 28, 489-514. Rosen, Sherwin, 1992, “Contracts and the Market for Executives”, in Werin, Lars and Hans Wijkander, eds., Contract economics, Cambridge, Mass. and Oxford; Blackwell, 181-211. Srinivasan, Aruna, and Larry D. Wall, 1992, "Cost Savings Associated with Bank Mergers", Federal Reserve Bank of Atlanta Working Paper 92-2, February.

Footnotes 1

Rhoades (1994) reports approximately 4,000 acquisitions between 1986 and 1994, including 119 where

both the acquiring and target banks had assets of more than $1 billion. 2

The connection between compensation and firm size has a long and consistent history. See Roberts

(1956), Ciscel and Carroll (1980) and Agarwal (1981) among others. For evidence on the pay-forperformance relationship in a cross-section of firms, see Murphy (1985, 1986) and Jensen and Murphy (1990). Results specific to the banking industry can be found in Barro and Barro (1990) and Houston and James (1992). 3

There is anecdotal evidence of a connection between CEO compensation and bank merger activity.

First Chicago NBD Corp.’s 1997 proxy statement attributes the 37% increase in CEO Verne Istock’s annual compensation to the “…successful merger of First Chicago and NBD.” Walter V. Shipley, CEO of Chase Manhattan Corp., received a “special merger bonus” of $5 million, payable in equal annual installments beginning in 1996. (“Chase Chairman Saw ’96 Income More Than Double”, The Wall Street Journal, March 31, 1997, page A4.) 4

As noted previously, these include Roberts (1956), Ciscel and Carroll (1980) and Agarwal (1981)

among others. 5

See Holderness and Sheehan (1988) and Kahn and Winton (1998) on the role of outside blockholders in

the monitoring process. 6

For example, NationsBank claimed that it would save almost $1 billion per year in costs when it

acquired Barnett Banks of Florida (The Wall Street Journal, September 2, 1997). Note that the expected savings are not always realized as banks frequently report unforeseen costs associated with mergers (for example, Banc One after merging with First USA in 1997, CoreStates after acquiring Meridian in 1996, and Fleet Financial after a series of acquisitions in 1995 all announced multimillion dollar additional costs associated with their mergers). 7

In summarizing 21 prior studies on stock price reaction to merger announcements, Rhoades (1996)

reports that “…the evidence regarding returns to bidders, as well as that regarding the net returns to bidders and targets combined, is too inconsistent to permit any clear conclusion." Papers that look at the

- 23 -

net effects include Baradwaj, Fraser, and Furtado (1990), de Cossio, Trifts, and Scanlon (1988) , Hannan and Wolken (1989), and Hawawini and Swary (1990). 8

As discussed below, the announcement effect for the acquirer may also reflect the issuance of

overpriced stock as discussed in a more general context in Myers and Majluf (1984). 9

Among the banks eliminated are four that failed during the first few years of the sample period.

10

Murphy (1998) shows that in a sample of firms, 83% had ten-year maturities and 95% of the options

were granted with an exercise price equal to the stock price at the time of the grant. 11

Most options are granted with a strike price equal to the stock price at the time of issue. Thus, since

stock prices trend upwards, most options will be in the money. Also, when stock prices fall significantly, Boards of Directors often rework existing options to lower the strike prices. 12

Although acquisitions of failed banks are excluded from our study, most of the mergers in 1988 were

takeovers of weak banks. 13

It would be nice if the difference in total compensation was statistically significantly different in the

two groups. One reason that it is not is that many banks did not give options grants to their CEOs every year. Option grants could cover performance in the past two or three years. Thus, a big option grant in 1986, the first year of the sample, or a lack of one in 1995, the last year of the sample, would skew the total compensation gain downward. These situations appear to happen in several of the banks in the sample, possibly adding noise to the total compensation gains. 14

All flow variables are for the year of the compensation observation, while all stock variables are as of

the end of the year of the observation.

Table 1 - Sample Banks and Descriptive Statistics A bank is in the sample if it was in the top 30 by asset size in any year between 1986 and 1995, and if there were five years of data between 1986 and 1995. There are 32 banks in the sample, and 298 bank-year observations. Cash compensation is defined as salary plus bonus. Total compensation is defined as cash compensation plus the value of stock options and restricted stock granted during the prior year. All dollar figures are 1995 dollars.

Panel A: Sample banks Banc One Bank of Boston Bank of New York Bankamerica Bankers Trust Barnett Banks Boatmen’s Bancshares Chase Manhattan Chemical Citicorp Comerica CoreStates First Chicago First Fidelity First Interstate First Union

Fleet Financial JP Morgan Key Corp. Manufacturer’s Hanover Mellon Bank National City Bank Nationsbank (NCNB) NBD Bancorp Norwest PNC Bank Security Pacific Shawmut National Society Corp. SunTrust Banks Wachovia Wells Fargo

Panel B: Descriptive statistics Mean

Median

Minimum

Maximum

64.4

44.0

11.8

304.0

Equity – market value

4.7

3.3

0.4

28.7

Equity – accounting value

4.0

3.0

0.8

20.8

Return on assets (percentage)

0.87

1.04

-2.45

2.44

Total compensation

2.82

2.19

0.34

16.37

Cash compensation

1.61

1.37

0.12

4.54

Cash compensation / total compensation

0.66

0.67

0.09

1.00

Total compensation / assets

56.1

48.0

6.3

206.2

Cash compensation / assets

34.5

31.0

3.6

139.4

104.6

66.1

2.7

763.8

Assets

($ billions) ($ billions) ($ billions)

($ millions) ($ millions)

($ / $ millions) ($ / $ millions)

Change in wealth from a 1% change in stock price ($ thousands)

Table 2 - Data on Bank Mergers Data on sample mergers by sample banks. A bank is in the sample if it was in the top 30 by asset size in any year between 1986 and 1995, and there were five years of data between 1986 and 1995. A megamerger is one where the target bank is at least 10 percent of the size of the acquiring bank prior to the merger. Announcement stock price reactions are the net percentage stock price change in the three-day window centered on the merger announcement day. The net change is the difference between the change in the acquirer's stock and the change in an index of bank stocks. All values are reported in the year of the merger, not the year of the announcement. All dollar values are 1995 dollars.

Year

Number of megamergers in the sample

Total assets of sample megamerger targets ($ billions)

Average assets of Average ratio Total assets sample of assets acquired in megamerger acquired in a megamergers / targets ($ megamerger total assets of all billions) banks in sample

Average All mergers: announcement Number of stock price banks making reaction for at least one megamergers acquisition

All mergers: Total assets acquired by sample banks ($ billions)

84 85 86 87 88 89 90 91 92 93 94 95

5 11 9 5 9 1 2 5 7 4 4 4

18.7 54.9 44.5 38.4 105.4 7.9 15.3 149.2 132.4 35.6 61.6 112.0

3.7 5.0 4.9 7.7 11.7 7.9 7.6 29.8 18.9 8.9 15.4 30.4

46.7% 45.2% 29.4% 21.0% 59.9% 40.3% 38.5% 46.3% 47.2% 25.8% 50.2% 63.1%

1.5% 4.0% 2.9% 2.1% 5.8% 0.4% 0.8% 8.5% 6.8% 1.7% 2.8% 5.1%

-1.5% -3.0% 0.2% -2.7% -4.4% -5.0% -7.7% -1.4% -1.1% -4.6% -1.6% -2.7%

8 11 17 12 17 9 8 9 12 17 17 16

30.7 55.8 64.8 48.3 116.4 9.1 30.2 155.0 166.4 71.7 100.3 146.1

All years

66

786.1

11.9

44.1%

3.6%

-2.4%

153

1006.0

Table 3 - High and Low-Merger Banks: Descriptive Statistics The sample banks are divided into high- and low-merger groups by the average level of merger activity over the sample period. There are sixteen banks in each group. The high-merger group has 156 bank-year observations while the low-merger bank group has 142 bank-year observations. The difference in the number of observations is because not all banks are in the sample for the entire period. Cash compensation is defined as salary plus bonus. Total compensation is defined as cash compensation plus the value of stock options and restricted stock granted during the prior year. All dollar figures are 1995 dollars.

Panel A. High–merger banks Banc One Bank of New York Bankamerica Boatmen’s Bancshares Chemical Comerica CoreStates First Union

Low–merger banks Fleet Financial Key Corp. National City Bank Nationsbank (NCNB) NBD Bancorp PNC Bank Society Corp. Wells Fargo

Bank of Boston * Bankers Trust * Barnett Banks Chase Manhattan * Citicorp * First Chicago * First Fidelity First Interstate

* – These banks had no megamergers during the sample period.

JP Morgan * Manufacturer’s Hanover * Mellon Bank Norwest Security Pacific Shawmut National SunTrust Banks Wachovia

Table 3 (cont.) Panel B. High-merger sample

Low-merger sample

Mean

Median

Min

Max

Mean

Median

Min

Max

Assets

57.2

37.5

11.8

304.0

72.3

46.4

18.6

276.8

Equity - market value

4.9

3.4

0.7

25.6

4.5

3.4

0.4

28.7

Return on assets (percentage)

1.04

1.08

-1.41

2.44

0.68

0.95

-2.45

1.79

Total compensation

2.61

2.06

0.66

14.06

3.06

2.27

0.33

16.37

Cash compensation

1.60

1.40

0.55

4.54

1.62

1.36

0.12

4.53

Cash compensation / total compensation Total compensation / assets

0.69

0.69

0.21

1.0

0.62

0.63

0.08

1.0

59.4

50.7

10.1

206.2

51.9

45.1

6.3

203.3

Cash compensation / assets

39.4

34.4

5.5

139.4

29.0

27.2

3.6

96.2

Change in firm value due to change in index (INDEX CHG, $ million / 1% change ) Change in firm value due to idiosyncratic change (FIRM CHG,

783.9

306.4

-1,695.7

8,702.6

626.8

298.1

-2,826.3

9,702.6

-129.0

-64.3

-3,636.7

2,669.4

-81.4

18.1

-7,246.5

4,572.7

10,200.8

4,001.4

0.0

109,121.1

2,285.8

0.0

0.0

16,962.8

5,270.7

2,334.0

-47,504.4

59,202.3

3,471.6

2,854.2

-47,012.1

74,202.6

55.59

56

42

67

57.5

58

48

66

0.278

0.212

0.017

0.983

0.332

0.280

0.048

1.006

($ billions) ($ billions)

($ millions) ($ millions)

($ / $ millions) ($ / $ millions)

$ million / 1% change)

Growth from megamergers (MEGA3, $ millions, years t-2 through t) Growth excluding megamergers (MGROW3, $ millions, years t-2 through t) Age of CEO shareholding (AGE, years) CEO shareholding (CEO SHR, % of total)

- 28 -

Table 4 – Compensation and merger activity The relation ship between merger activity and compensation. Cash compensation is defined as salary plus bonus. Total compensation is defined as cash compensation plus the value of stock options and restricted stock granted during the prior year. The sample banks are divided into high- and low-merger groups by the average level of merger activity over the sample period. There are sixteen banks in each group. Compensation gains are the annual increase in compensation during the sample period, excluding a CEO’s first year in office. Aggregate rate is the annual rate for nine years. The sample period is 1986 - 1995.

Full sample

Annual rate of change in total compensation 10.3%

Aggregate rate of change in total compensation 141.6%

Annual rate of change in cash compensation 9.9%

Aggregate rate of change in cash compensation 133.0%

High-merger group

10.8%

151.6%

12.0%

177.3%

Low-merger group

9.8%

131.9%

7.7%

95.2%

Difference

1.0%

19.7%

4.3%*

82.1%*

* -- High- and low-merger groups are significantly different from each other at the 10% level.

Table 5 – Regression Results Regressions of compensation measures against changes in firm value due to stock price changes, changes in asset size, and other control variables. Cash compensation is defined as salary plus bonus. Total compensation is defined as cash compensation plus the value of stock options and restricted stock granted during the prior year. The sample period is 1986-1995. Year dummies are not shown. For all coefficients, t-statistics are shown in parentheses. All standards errors are adjusted using White’s correction. All regressions have 298 observations.

(1) Total Compensation ($ millions)

(2) Total Compensation ($ millions)

(3) Cash Compensation ($ millions)

(4) Cash Compensation ($ millions)

Total assets, year t-3 (RTA3,$ billions)

36.86 (1.82)

37.34 (1.86)

15.51 (1.08)

15.98 (1.13)

Assets acquired in megamergers, years t-2 - t (MEGA3, $ billions)

53.55 (2.61)

Dependent Variable:

Assets acquired in all mergers, years t-2 – t (MRG3, $ billions), Non-megamerger growth, year t-2 – t (MGROW3, $ billions)

26.15 (1.86) 58.84 (2.86)

30.22 (1.50)

Non-merger growth, year t-2 t (GROW3, $ billions)

31.65 (2.26) 12.26 (0.88)

23.70 (1.21)

6.15 (0.45)

Change in firm value due to change in index (INDEX CHG, $ millions)

951.57 (4.72)

952.46 (4.74)

284.33 (2.47)

286.64 (2.52)

Change in firm value due to idiosyncratic change (FIRM CHG, $ millions)

535.83 (2.92)

545.69 (3.01)

156.41 (1.50)

161.99 (1.58)

Return on assets (ROA, percentage * 1000)

10.74 (2.85)

10.87 (2.90)

10.21 (4.49)

10.40 (4.58)

Age (AGE, years * 1000)

-0.18 (0.53)

-0.20 (0.59)

0.05 (0.21)

0.03 (0.13)

CEO shareholding (CEO SHR, percent of total equity * 1,000,000)

67.27 (5.97)

67.46 (6.04)

31.96 (4.95)

32.22 (5.06)

Adj. R-squared

0.312

0.321

0.217

0.235

Table 6 - Regression Results: Estimated changes in compensation and wealth. Estimated changes in CEO compensation and wealth based on the reported results in Table 5, column 1. The results are estimated for each acquisition by 53.55 * assets acquired in merger + 535.83 * change in stock equity value. The change in stock equity value is calculated two ways. First, we find the change in equity value of the acquiring bank in the three-day window around the announcement day. Second, we use the average percentage change in equity value of banks in the sample in the three-day window around an announcement day. This is multiplied by the equity value of the acquiring bank. When indicated, these values are divided by the compensation of the acquiring CEO. The sample period is 1986 – 1995. Panel A. Changes in compensation.

Mean Median 25 percentile 75 percentile

Based on actual CAR Total compensation Cash compensation $986,235 38.20% $487,173 28.15% $407,804 26.63% $202,700 20.03% $245,146 10.92% $136,030 11.26% $880,700 40.84% $418,167 39.20%

Based on average CAR Total compensation Cash compensation $944,295 37.60% $474,931 27.64% $461,108 26.40% $246,725 19.75% $303,595 13.60% $158,424 11.19% $836,658 38.90% $415,999 34.22%

Note: 98.11% of CEOs had an increase in estimated total compensation using the actual CAR. 100.00% of CEOs had an increase in estimated total compensation using the average CAR. Panel B. Changes in wealth.

Mean Median 25 percentile 75 percentile

Based on actual CAR Based on average CAR Total compensation Total compensation + wealth change + wealth change Wealth change Wealth change -$29,264 $956,971 31.78% -$228,359 $715,936 29.36% -$110,704 $286,494 16.12% -$155,728 $279,870 17.98% -$413,182 $36,219 1.90% -$281,313 $103,584 6.17% -$1,597 $786,475 43.62% -$79,721 $710,556 33.83%

Note: 77.36% of CEOs had an increase in estimated total compensation + wealth change using the actual CAR. 88.68% of CEOs had an increase in estimated total compensation + wealth change using the average CAR.

- 31 -

Table 7 - Regression Results: High-merger versus low-merger banks Regressions of compensation measures against changes in firm value due to stock price changes, changes in asset size, and other control variables. Cash compensation is defined as salary plus bonus. Total compensation is defined as cash compensation plus the value of stock options and restricted stock granted during the prior year. The sample period is 1986 - 1995. The sample banks are divided into high- and low-merger banks by the average level of merger activity over the sample period. There are sixteen banks in each group. The low-merger with at least one megamerger group contains the nine low-merger banks with at least one megamerger. Year dummies are not shown. For all coefficients, t-statistics are shown in parentheses. All standards errors are adjusted using White's correction. Dependent Variable:

Total Compensation ($ millions)

Cash compensation ($ millions)

(1)

(2)

(4)

(5)

High-merger banks

Low-merger banks

High-merger banks

Low-merger banks

Total assets, year t-3 (RTA3,$ billions)

18.16 (0.65)

70.00 (2.59)

-12.09 (0.61)

24.03 (1.65)

Assets acquired in megamergers, years t-2 - t (MEGA3, $ billions)

21.80 (0.78)

110.72 (3.57)

-7.66 (0.40)

39.99 (2.28)

Non-megamerger growth, year t-2 – t (MGROW3, $ billions)

17.26 (0.70)

66.01 (2.09)

-1.34 (0.07)

10.91 (0.64)

Change in firm value due to change in index (INDEX CHG, $ millions)

752.72 (3.02)

1534.65 (5.03)

432.18 (2.55)

248.63 (1.32)

Change in firm value due to idiosyncratic change (FIRM CHG, $ millions)

291.90 (1.15)

1024.70 (4.56)

29.12 (0.20)

278.44 (1.89)

Return on assets (ROA, percentage * 1000)

16.76 (3.42)

0.72 (0.16)

13.92 (3.79)

5.19 (2.16)

Age (AGE, years * 1000)

0.39 (0.85)

-1.03 (-2.22)

0.60 (2.01)

-0.18 (0.74)

CEO shareholding (CEO SHR, percent of total equity * 1,000,000)

65.57 (4.47)

70.86 (3.85)

38.79 (3.68)

27.07 (3.41)

Adj. R-squared Number of observations

0.310 156

0.365 142

0.248 152

0.224 146

- 32 -

Table 8 - Estimated changes in compensation and wealth for the high- and low-merger banks. Estimated changes in CEO compensation and wealth based on the reported results in Table 7, columns 1 and 2. The reported results are reported on a per year basis rather than a per merger basis to allow a direct comparison of the aggregate gains at high- and low-merger banks. The results are estimated for each acquisition by 21.80 * assets acquired in merger + 291.90 * change in stock equity value for the high-merger banks and 110.72 * assets acquired in merger + 1024.70 * change in stock equity value for the low-merger banks. The change in stock equity value is calculated two ways. First, we find the change in equity value of the acquiring bank in the three-day window around the announcement day. Second, we use the average percentage change in equity value of banks in the entire sample in the three-day window around an announcement day. This is multiplied by the equity value of the acquiring bank. The sample period is 1986 – 1995.

Panel A. Change per year in compensation and wealth at high-merger banks from acquisitions.

Mean Median 25 percentile 75 percentile

Based on actual CAR Total compensation + Total compensation wealth change $121,592 $139,761 $88,106 $67,338 $38,433 -$40,049 $178,212 $175,504

Based on average CAR Total compensation + Total compensation wealth change $112,510 $50,491 $89,324 $33,831 $45,103 -$3,975 $151,144 $77,927

Note: 68.75% of CEOs had an increase in estimated total compensation + wealth change using the actual CAR. 75.00% of CEOs had an increase in estimated total compensation + wealth change using the average CAR.

Panel B. Change per year in compensation and wealth at low-merger banks from acquisitions.

Mean Median 25 percentile 75 percentile

Based on actual CAR Total compensation + Total compensation wealth change $70,575 $42,171 $70,438 $0 $0 $0 $132,364 $129,905

Based on average CAR Total compensation + Total compensation wealth change $26,573 $62,433 $26,767 $45,768 $0 $0 $47,954 $101,338

Note: 70.00% of CEOs at banks with mergers had an increase in estimated total compensation + wealth change using the actual CAR. 100.00% of CEOs at banks with mergers had an increase in estimated total compensation + wealth change using the average CAR.

Table 9 - Logit Results for Probability of a Merger Announcement The dependent variable is 1 if a merger announcement is made in year t + 1 and 0 otherwise. The independent variables includes the form of compensation received by a CEO and controls for changes in stock price (expressed as the percentage change in the bank stock index and the percentage change in a bank's stock value not due to changes in the bank stock index), merger activity, a CEO age dummy, the fraction of the board that are bank insiders, the log of total assets, and year dummies (not shown). Cash compensation is defined as salary plus bonus. Total compensation is defined as cash compensation plus the value of stock options and restricted stock granted during the prior year. p-values are in parentheses. All regressions have 295 observations. Dependent variable: 1 if an merger announcement is made during year t + 1 and 0 otherwise Ratio of cash compensation to total compensation

(1)

(2)

2.77 (0.06)

Dollar value of stock ownership ($ millions)

(3)

(4)

3.50 (0.03) -0.16 (0.05)

-0.21 (0.02)

Dollar value of new stock compensation ($ millions)

-0.99 (0.06)

Average number of mergers announced by the bank

15.58 (0.00)

16.87 (0.00)

16.36 (0.00)

15.60 (0.00)

Assets acquired in mergers as a fraction of total assets, years t-2 to t (percentage)

-4.78 (0.00)

-4.77 (0.00)

-4.79 (0.00)

-4.46 (0.00)

Growth above bank stock index, year t-2 – t (% return)

0.04 (0.09)

0.04 (0.13)

0.04 (0.10)

0.05 (0.05)

Market-to-book ratio of the bank, end of year t

0.02 (0.11)

0.07 (0.01)

0.09 (0.01)

0.03 (0.07)

CEO 50 years old and younger dummy

1.31 (0.04)

1.19 (0.06)

1.44 (0.03)

1.13 (0.07)

Fraction of board that is insiders (percentage)

-0.06 (0.08)

-0.04 (0.23)

-0.04 (0.26)

-0.06 (0.09)

Log of real total assets in year t

-0.22 (0.84)

1.22 (0.41)

2.66 (0.11)

0.50 (0.68)

Pseudo R-squared

0.472

0.471

0.449

0.458

- 34 -

Figure 1. Evidence of the changes in banking, 1985 - 1995.

12000 11000

56.0% 54.0% 52.0% 50.0%

10000 9000 8000

48.0% 46.0% 44.0%

7000 6000

Number of banks

58.0%

19 85 19 86 19 87 19 88 19 89 19 90 19 91 19 92 19 93 19 94 19 95

Banks' share of nonfinancial corporate debt

60.0%

Banks' share

Number of banks