Delegation and incentive compensation.

I. INTRODUCTION

Because top management cannot make all the firm's operational decisions, it faces two critical organizational design choices: (1) how much authority to delegate to lower-level employees, and (2) how to design incentive compensation to ensure that these employees do not misuse

their discretion (Milgrom and Roberts 1992, Chapter 16; Stiglitz 1994, Chapter 9; Brickley et al. 1996, Chapter 8). A large body of theoretical accounting research has formalized the joint nature of the firm's delegation and incentive compensation choices (Melumad and Reichelstein 1987; Melumad et al. 1992; Baiman and Rajan 1995; Bushman et al. 2000). Responsibility accounting, a fundamental concept in management accounting, also stresses the link between delegation and incentive compensation, suggesting that lower-level employees' incentives should be closely related to their responsibilities (Horngren et al. 2000, 195-196).

Prior research, however, provides little empirical evidence on how top management balances the delegation of authority and the extent of incentive compensation awarded to lower-level managers. Holthausen et al. (1995) and Bushman et al. (1995) examine incentive compensation for divisional managers, but they do not control for the authority delegated to these managers. Baiman et al. (1995) incorporate both delegation and compensation choices in their study, but examine these two choices separately as two different functions of firm characteristics.

The difficulty in acquiring delegation data has also hampered empirical investigation of the joint nature of firms' delegation and incentive compensation choices. For example, Baiman et al. (1995) can only construct a binary proxy for delegation. I capitalize on a unique cross-sectional database of retail banks with relatively detailed information on delegation and incentive compensation choices for branch managers. This database stems from a survey the Wharton Financial Institutions Center (WFIC) conducted in the mid-1990s to understand how banks structure and manage their retail operations.

I use prior theory to develop a simultaneous model of the retail banks' delegation and incentive compensation choices for branch managers. First, the results indicate that high-growth, volatile, and innovative banks delegate more authority to their branch managers than do more stable banks. Second, I find that the extent of the branch manager's incentive compensation increases with the extent of authority delegated to the manager, after controlling for several firm factors. In fact, firm characteristics such as growth, volatility, and innovation play no incremental role in explaining the extent of incentive compensation, after controlling for the extent of delegation. On the other hand, in contrast with principal-agent theory, I find no evidence that the extent of incentive compensation plays a significant role in explaining the extent of delegation. However, I cannot rule out the possibility that this simply reflects the statistical limitations of analyses based on a relatively small sample of 100 banks.

Prior divisional manager compensation studies' failure to control for the extent of delegation can potentially explain some of their puzzling results. In their study of divisional manager compensation, Baiman et al. (1995) predict a positive association between CEO expertise and divisional manager incentive compensation. They argue that an expert CEO can more effectively interpret the divisional performance measures. This lowers the risk to the divisional manager of bonuses based on divisional performance measures, which in turn reduces the cost to the firm of instituting strong incentive pay. However, in contrast to this prediction, Baiman et al. (1995, 224) find that greater CEO division expertise is associated with less powerful incentive compensation for divisional managers. The delegation perspective provides an explanation for this result: CEOs with divisional expertise are less likely to delegate decisions to divisional managers, thus reducing the need to provide the divisional managers with incentive compensation. Indeed, Baiman et al. (1995, 225) find that expert CEOs are less likely to delegate, but do not incorporate this association in explaining the incentive compensation choice.

The remainder of the paper is organized as follows. Section II provides a theoretical discussion of the firm's delegation and incentive compensation choices, based on prior literature. Section III discusses the data set. Section IV develops the empirical measures and the model specification. Section V discusses the results, and Section VI concludes.

II. THEORY

Information required for operational decisions changes rapidly in firms operating in high-growth and uncertain environments. Prior theory suggests that the costs for top management to acquire this information are high in volatile and high-growth firms; consequently, top management delegates authority to divisional managers who have closer access to the necessary information. By delegating authority to these divisional managers, the firm can use this information effectively, while avoiding the costs of collecting and transmitting all information to top management (Melumad and Reichelstein 1987; Kirby 1987; Jensen and Meckling 1992; Bushman et al. 2000; Prendergast 2000).

Although delegation reduces the cost of information gathering, it may create additional costs if divisional managers misuse their authority. To mitigate this moral hazard problem, firms develop incentive compensation linking divisional manager compensation to divisional performance. However, Prendergast (2000) argues that in rapidly changing environments, divisional performance measures are likely to fluctuate for reasons beyond the divisional manager's control. Principal-agent theory indicates that incentives based on noisy performance measures impose risk on the divisional manager, for which the firm must compensate (Kreps 1990, Chapter 16). This compensation, called the risk premium, is the incentive-related cost of delegation, which top management trades off against the benefits of delegation. Thus, top management makes the delegation choice based on both the incentive compensation choice and the nature of the firm's operating environment:

(1) extent of delegation

= f(extent of incentive compensation, firm's operating environment)

Because incentive compensation is a cost of delegation, I expect incentive compensation to have a negative coefficient in Equation (1).

The argument outlined above also suggests that incentive compensation is a function of the level of delegation. Further, holding delegation constant, principal-agent theory posits that the more risky the operating environment, the weaker the incentive compensation (Prendergast 2000; Kreps 1990, Chapter 16). Performance measures that form the basis of incentive compensation are more likely to fluctuate in such an environment for reasons beyond the divisional manager's control, and therefore impose greater risk on the divisional manager. Compensating the divisional manager for this risk is costly to the firm. In equilibrium, the firm lowers this cost by offering a weaker incentive contract. Thus, top management makes incentive compensation decisions based not only on the delegation choice, but also on the nature of the firm's operating environment:

(2) extent of incentive compensation

= f(extent of delegation, firms' operating environment)

I expect managers with more authority to receive more incentive compensation, so I expect the delegation coefficient to be positive in Equation (2). Equations (1) and (2) form the basis of this study's simultaneous model of delegation and incentive compensation choices. (1)

Several organizational theories view delegation and incentive compensation as the two key choices facing top management (Milgrom and Roberts 1992, Chapter 16; Stiglitz 1994, Chapter 9; Brickley et al. 1996, Chapter 8). However, organizational design is not limited to these two choices, and thus this study's analysis is a simplified partial equilibrium analysis that takes other choices as given. In particular, like Baiman et al. (1995), I do not examine an important aspect of incentive compensation: the choice of specific performance measures in divisional manager incentive contracts (see Bushman et al. 1995; Brickley et al. 1996, Chapter 8). However, this does not appear to be a serious omission, given this study's exclusive focus on retail banking. Banks' widespread use of branch earnings as a performance measure precludes the need to model explicitly the performance measure choice. For example, Ittner et al. (2001) find that branch earnings are a key performance measure for branch manager bonuses, and Ittner and Larcker (1998) find some evidence that branch earnings capture the branch manager's actions in a timelier manner relative to other performance measures such as customer satisfaction. Eighty-nine percent of this study's sample uses branch earnings as a basis for branch manager bonuses (though the data on the proportion of bonus based on branch earnings are not available).

III. DATA

This study uses data from two sources. The first source is the Wharton Financial Institutions Center's (WFIC) 1994 survey of retail banks. This survey provides information on the organizational practices in branch banking, the most prominent retail distribution channel at the time (Bank Administration Institute 1998). The banking industry also experienced a rapidly changing environment in the 1990s, with deregulation and the emergence of other retail financial institutions such as mutual funds. Within this environment, banks pursued various organizational strategies in response to external changes. Therefore, despite a relatively homogeneous production function, the sample provides an interesting cross-section of organizational structures and strategies.

The WFIC survey respondents include senior executives of retail banks and branch managers of the head branches. The survey was completed in two phases: first, the WFIC research team approached the 70 largest U.S. bank holding companies. From this group, the team secured the participation of 47 bank holding companies, yielding a sample of 64 banks (seven bank holding companies provided information on two or more banks). The team then contacted the 265 next-largest bank holding companies by mail. From this group, 64 bank holding companies agreed to participate, yielding 71 more banks, and resulting in a total sample of 135 banks.

The second source of data is the bank call reports filed with the Federal Deposit Insurance Corporation (FDIC). The call reports contain quarterly balance sheets, income statements, and other financial data that regulators use to monitor the banks' financial conditions. To ensure that the banks in the final sample have data available from both sources, I match the banks in the WFIC survey data with those in the call reports by name and city, to the extent possible. The relevant data are all available for only 100 banks, which comprise this study's sample.

The combined assets of the banks in the final sample sum to about 22 percent of the total assets of the approximately 10,000 banks listed in the FDIC database. The sample's 1994 capitalization ratio and average net income as a percentage of assets are 9 percent and 1.26 percent, respectively, indicating that the sample banks are, on average, financially healthy.

Although the WFIC survey contacted several employees per bank, it contacted only one employee per bank for each question in the survey (the Appendix provides the identity of the respondent for each survey question). Thus the survey data provide only one response per bank, even though banks typically have multiple branches. Though a limitation, research on divisional manager compensation research often uses typical or average division measures (Holthausen et al. 1995, 294).

IV. VARIABLE DEFINITIONS AND MODEL SPECIFICATION

This section first describes the measures in the model of the link between delegation and incentive compensation. It then describes the model. Several measures in the model, such as the extent of delegation of decision rights and the nature of the firm's operating environment, are difficult to measure empirically (Jensen 1983). To address this issue, the empirical constructs are based on prior literature, and when possible, measured with multiple variables. The Appendix provides data-specific details on these measures.

Delegation

To measure the extent of delegation from top management to the branch, I use survey items that explore the amount of branch manager authority relative to top management authority in various branch operations. Specifically, I use the following survey questions (details appear in the Appendix):

1) Discretion available to the branch manager in hiring tellers.

2) Discretion available to the branch manager in awarding promotions in the branch.

3) Discretion available to the branch manager in determining the hours the branch is open.

4) Discretion available to the branch manager in changing the process for selling new investment products.

In particular, I choose investment products in the fourth question because it reflects the amount of authority the branch manager has over a complex and important retail product.

To assess whether these items reflect a common underlying delegation construct and to reduce the dimensionality of the measure, I conduct a factor analysis. This factor analysis yields one factor with an eigenvalue greater than unity, on which all four items load with weights greater than 0.45. This factor, described in detail in the Appendix, is:

DELEGATE: The standardized aggregated sum of the branch manager's authority in hiring, promoting, setting hours, and changing selling processes.

Although this is an aggregate measure, prior divisional-manager compensation studies have used coarser measures of delegation. For instance, Baiman et al. (1995, 228) use a binary variable to represent whether a division has control over its core functions.

Incentive Compensation

Baker et al. (1994) argue that incentive compensation is the primary performance-based reward scheme in the firm, and that supervisors identify employees who consistently earn high bonuses as talented and thus promote them to the next level. I use the following WFIC survey question to measure the amount of the branch manager's incentive compensation (details appear in the Appendix):

INCENTIVE: The proportion of the typical branch manager's pay that is bonus-based.

INCENTIVE is a categorical measure of whether the typical magnitude of the bonus lies within certain ranges, e.g., "1-6 percent of total pay," or "7-10 percent," etc., with the highest category being "25 percent and above." Theory suggests that managers with more authority have the potential to earn more incentive compensation. Since the WFIC survey question does not specify whether respondents should indicate potential or actual compensation, INCENTIVE may be a noisy indicator of performance compensation. However, previous research suggests that expected and actual bonuses are very close (Holthausen et al. 1995; Baiman et al. 1995, note 17).

Firm's Operating Environment

To measure the extent to which the firm operates in rapidly changing and uncertain environments, I use three measures: firm growth, volatility of earnings, and the firm's strategic emphasis on innovation. Theory suggests that these factors affect the extent to which top management delegates. If the firm is growing or changing rapidly, then centralized decision making requires top management to track information continuously, which is costly. Delegation of authority to branch managers, who typically have such information due to close, ongoing contacts with customers, becomes an attractive alternative (Baiman and Rajah 1995; Kaplan and Atkinson 1998, Chapter 7). Demsetz (1992) and Brickley et al. (1996, Chapter 8) argue that innovative firms emphasizing new products and new market segments require information such as customer preferences, local opportunities, and lower-level employee adaptability to new products. The intangible nature of this information increases its transmission and interpretation costs across the firm, making delegation an attractive alternative. This study therefore includes the three firm characteristics, growth, volatility, and strategic emphasis on innovation, in the delegation equation.

In addition to affecting delegation, the three firm characteristics directly affect incentive compensation. Theory suggests that, holding all else equal, greater risk or more noise in performance measures imposes more risk on the divisional manager, and therefore reduces the extent to which the firm will use incentive compensation (Prendergast 2000). Prior studies suggest that high-growth and volatile firms' performance measures fluctuate more for exogenous reasons beyond the manager's control, thus imposing more risk on the divisional manager (e.g., Lambert and Larcker 1987). Furthermore, Holthausen et al. (1995) argue that it is difficult to measure accurately the outcomes of managerial actions in firms that place strategic emphasis on innovation of new products and services. This study therefore includes the three firm characteristics in the incentive compensation equation as well.

I next turn to operational measures of growth, volatility, and innovation. To measure firm growth, I use insured deposit growth, since the American Bankers Association (1998) survey indicates that two out of three customers initiate banking relationships with deposits. Given that the WFIC survey took place in December 1994, I measure growth from June 1994 through June 1995 to cover growth prior to December 1994 and anticipated future growth. The firm growth measure is:

GROWTH: The growth in insured deposits from June 1994-June 1995.

Because earnings comprise one of the most important performance measures for branch manager compensation (Ittner et al. 2001), I use the volatility of earnings as a measure of volatility facing the branch manager. Due to unavailability of individual branch earnings data, I use overall bank earnings (scaled by assets to control for heteroskedasticity) to compute volatility. This measure is:

STDROA: Standard deviation of the bank's net income scaled by assets from 1990-1994. (2)

To measure innovation, I use the WFIC survey item on the firm's strategic emphasis on innovation. This survey measure is:

INNOV: The extent to which the firm emphasizes innovation in the design and delivery of products and services.

Finally, since firm size is associated with all critical organizational design choices, I control for the bank's retail division size, using the log of deposits:

SIZE: The log of insured deposits. (3)

Exogenous Variable for the Delegation Equation

To ensure identification, the simultaneous estimation of Equations (1) and (2) requires exogenous variables for each equation. Although such variables should directly affect only the relevant dependent variable, one can often make a plausible case that a given exogenous variable directly affects all dependent variables (Ittner and Larcker 2001). To mitigate this problem, this study's selection of identifying variables is grounded in prior literature.

I use the number of acquisitions the bank made as the exogenous variable affecting delegation. Acquisitions should reduce delegation, as a successful acquisition requires top-level streamlining and integration of systems, employees, and customers (Ashkenas et al. 1998; Darnell 1999; Carey 2000). Such control can curtail lower-level employees' decision rights. (4) In contrast to this direct effect on delegation, acquisitions are unlikely to affect the risk or noise in performance measures, especially after controlling for firm size, growth, volatility, and strategy. Thus, acquisitions should affect incentive compensation only indirectly, through the delegation effect?

There is typically a delay between the announcement of an acquisition and its occurrence. To allow for both past and anticipated acquisitions, I count the number of acquisitions the bank made from June 1994 through June 1995. I also control for the size of the acquisitions:

ACQUIRE: The sum of the number of acquisitions the bank made from June 1994 through June 1995, adjusted for size. Following the FDIC categorization, if a particular acquisition increases assets by less than 25 percent, then it is counted as 1; otherwise, it is counted as 2. I then sum all the bank's acquisitions over the June 1994-June 1995 period.

Thus, I estimate the following empirical model for the delegation choice (predicted signs in parentheses):

(3)                       (-)                   (+)
DELEGATE = [a.sub.1] + [a.sub.2] INCENTIVE + [a.sub.3] GROWTH +

    (+)               (+)               (?)              (-)
[a.sub.4] STDROA + [a.sub.5] INNOV + [a.sub.6] SIZE + [a.sub.7]

ACQUIRE + [[epsilon].sub.1].

In this equation, this first regressor, INCENTIVE, should have a negative coefficient as stronger incentives imply a higher risk premium. This premium makes delegation more costly for the firm and, ceteris paribus, lowers the equilibrium level of delegation (Prendergast 2000). I expect the next three coefficients on the right side of the equation to be positive, as greater growth, volatility, and innovation require more delegation. I do not predict the direction of the size effect, as firm size can proxy for several constructs. (6) Finally, I expect the acquisition coefficient to be negative, as acquisitions require more centralized decision making.

Exogenous Variables for the Incentive Compensation Equation

I use proxies for the branch manager's risk aversion as exogenous variables for the incentive Equation (2). Since incentive-related uncertainties are more costly for divisional managers with higher risk aversion, the principal-agent model predicts that, ceteris paribus, firms offer weaker incentive contracts to divisional managers who are more risk-averse. Furthermore, Kreps (1990, Chapter 16) suggests that the divisional manager's risk aversion does not directly affect the delegation choice in Equation (1), because the delegation costs to the firm arise not from the divisional manager's risk characteristics, per se, but from the payment of incentive compensation wages, which I include as a regressor in the delegation equation.

Prior research finds that the manager's education and experience are associated with the manager's risk aversion. Examining the mutual fund industry, Chevalier and Ellison (1999a, 1999b) find that the risks managers take in their operating decisions increase with manager education and age (a proxy for experience). I, therefore, measure the branch manager's risk aversion by his education and experience with other banks. Educated and experienced branch managers are likely to be wealthier and more confident of their abilities, and thus more tolerant of the financial risk in performance-based rewards. The specific risk aversion measures, described in detail in the Appendix, are:

EDUCATE: Level of the branch manager's education.

EXPERIENCE: A binary variable that measures whether the bank manager has worked for other banks. (7)

If educated branch managers are awarded more decision rights, then one could argue that EDUCATE also belongs in the delegation Equation (1). However, since such managers demand adequate compensation for their talent, and incentive compensation is a regressor in the delegation equation, the effect of education on delegation is to some extent already reflected in the delegation equation. Section V documents the robustness of the results to this specification assumption by including EDUCATE as an additional regressor in the. delegation equation.

In sum, including the exogenous variables leads to the following incentive compensation equation (predicted signs in parentheses):

                            (+)                  (-)
(4) INCENTIVE = [b.sub.1] + [b.sub.2] DELEGATE + [b.sub.3] GROWTH +

(-)                (-)               (?)              (+)
[b.sub.4] STDROA + [b.sub.5] INNOV + [b.sub.6] SIZE + [b.sub.7] EDUCATE

  (+)
+ [b.sub.8] [EXPERIENCE.sub.2]

In Equation (4), the first regressor, DELEGATE, should have a positive coefficient, as branch managers who exercise greater authority should receive more incentive compensation. The next three variables should have negative coefficients, as, ceteris paribus, high-growth, volatile, and innovative banks are more likely to have noisy performance measures that fluctuate for reasons beyond the branch manager's control. Such noise imposes risk on the branch manager, which the bank responds to by attenuating the extent of incentive compensation. As with the delegation equation, I do not predict the sign of the SIZE coefficient, because firm size can proxy for several constructs. Finally, both EDUCATE and EXPERIENCE should have positive coefficients, as they proxy for risk tolerance, which should be positively related to the extent of incentive compensation.

V. RESULTS

Table 1 presents descriptive statistics for all the measures in the two equations. The survey responses for organizational design measures such as the extent of incentive compensation, the extent of delegation, and the level of firm innovation span the entire set of feasible responses, so these measures have variation. The average incentive compensation comprises 7-10 percent of branch manager's pay.(8) The mean firm is somewhat innovative, and has a growth rate of 4.83 percent. The mean number of acquisitions in 0.49. A third of the banks in the sample made acquisitions, suggesting that the mean acquisition figure is not driven by a few banks. Finally 44 percent of the branch managers have worked in other banks, and the mean branch manager is college-educated.

Table 2 presents the correlations among the variables. Banks delegating more decision rights to branch managers also use more incentive compensation ([rho] = 0.19, significant at the 10 percent two-tailed level). As predicted, delegation is significantly higher in high-growth, volatile, and innovative banks ([rho] = 0.18, 0.19, and 0.27, respectively, all significant at the 10 percent two-tailed level). However, incentive compensation is not significantly associated with firm growth, volatility, or innovation.

The results in Table 2 also show that the relations between the endogenous variables and the identifying exogenous variables are as expected. The extent of delegation is significantly negatively related to acquisitions ([rho] = -0.17), while incentive compensation is significantly positively related to branch manager education and experience ([rho] = 0.36 and 0.19, respectively). The identifying exogenous variables are significantly correlated with only their respective endogenous variables and not the other endogenous variable, reinforcing the identifying variable selection.

Table 3 shows the result of the two-stage least squares (2SLS) estimation of Equations (3) and (4). The first-stage adjusted R2 is 17 percent for the delegation equation, and 19 percent for the incentive compensation equation, similar to that of Holthausen et al. (1995, 300). (9)

The first results column of Table 3 indicates that high-growth, volatile, and innovative banks are likely to delegate more authority (at the 10 percent two-tailed level), as predicted. The identifying variable ACQUIRE is significantly negatively associated with delegation (coefficient = -0.298, t-statistic = -2.23), consistent with the argument that banks engaging in acquisitions are more likely to centralize operations.

Incentive compensation, however, is not a significant predictor of delegation. Including EDUCATE (and, alternatively, EXPERIENCE) as a regressor in the delegation equation to control for any incremental education or experience effect on delegation does not change this result. The insignificance of incentive compensation is inconsistent with principal-agent theory, which predicts that incentive compensation is a major cost of delegation. However, an alternative explanation for the insignificance is that the noise in the first-stage estimation of the incentive compensation predictor, coupled with the relatively small sample size of 100 banks, reduces the power of the test.

Turning to the incentive equation, the extent of delegation is significantly positively associated with the extent of incentive compensation (coefficient = 1.145, t-statistic = 2.44). The Hausman endogeneity test shows significant endogeneity at the 10 percent level, thus justifying delegation as an endogenous choice variable. However, firm growth, volatility, and innovation are not significant individually (or jointly--the joint F-statistic is also insignificant at the 10 percent level). Since these measures proxy for the effect of risk on incentive compensation, the lack of significance may reflect either low risk variation in the sample or an inability of the measures to capture incentive compensation-related risk. (10) Finally, the identifying variable EDUCATE is highly significant, consistent with the argument that, holding delegation constant, a more educated (and presumably less risk-averse) branch manager will receive more incentive compensation in equilibrium.

In summary, the findings indicate that high-growth, volatile, and innovative banks delegate more authority to branch managers, and managers with more authority receive more incentive-based pay. These results provide some of the first empirical evidence on the link between the delegation and incentive compensation choices for lower-level managers in a firm.

The inferences are robust to several sensitivity tests. First, I repeat the tests using a finer measure of ACQUIRE, coding each merger as 1, 2, 3, or 4, depending on whether the merger increases the assets by 0-25 percent, 25-50 percent, 50-75 percent, or more than 75 percent. The results are virtually identical. Second, I test other measures of bank size. Including the number of branches as a regressor does not change the results. Holthausen et al. (1995) and Baiman et al. (1995) argue that division size also affects incentive compensation. I thus include average branch size (measured as firm size divided by the number of branches) as an additional regressor. This inclusion does not affect the results, nor does the inclusion of alternative growth measures such as loan growth. Finally, prior research argues that earnings and capital ratios affect top management's operating, financing, and accounting choices (Collins et al. 1995; Beatty et al. 1995). However, including ROA and the capital ratio as additional control variables does not change the inferences.

VI. CONCLUSIONS

Two key organizational design choices facing top management are: delegation of authority to lower-level managers, and the provision of incentive compensation to ensure that these employees do not misuse their discretion. This study examines empirically the link between delegation and incentive compensation for branch managers in retail banks. A simultaneous equation model of these two choices indicates that high-growth, volatile, and innovative banks delegate more authority to branch managers. In turn, managers with more authority receive more incentive-based pay. However, inconsistent with principal-agent theory (which argues that incentive compensation is a major cost of delegation), incentive pay does not affect the firm's delegation choice.

This study's findings are subject to several important caveats. First, no empirical study can simultaneously model all organizational design choices. For example, this study abstracts from the endogeneity of firm characteristics such as innovation. It also abstracts from other reasons for awarding incentive compensation (e.g., mechanism to hire good managers). Therefore, this study is a partial equilibrium analysis. Second, organizational design variables such as delegation, the nature of the firm's operating environment, and managers' risk aversion are measured with error. Third, this study uses observations from firms in a single industry polled at the same time. Cross-sectional dependencies are therefore a significant concern. Fourth, the isolation of exogenous variables to identify the simultaneous equation is difficult, and the empirical model is likely misspecified.

Despite these limitations, this study makes an important contribution to the empirical incentive compensation literature. Although accounting theory has repeatedly emphasized the interrelated nature of delegation and incentive compensation (Melumad and Reichelstein 1987; Melumad et al. 1992; Milgrom and Roberts 1992, Chapter 16; Baiman and Rajan 1995; Bushman et al. 2000), the extensive empirical incentive compensation literature contains, to the best of my knowledge, no direct evidence on this interrelation. This study provides some of the first empirical evidence on the joint nature of the film's delegation and incentive compensation choices for lower-level managers.

APPENDIX VARIABLE DEFINITIONS

Endogenous Variables

DELEGATE results from a factor analysis of the following four Wharton Financial Institutions Center's (WFIC) 1994 survey questions to the office of the senior executive of the retail bank: (11)

1. To what extent do the branch managers have a say in the hiring of tellers in the branch?

(Scale 1 to 7, as described below)

2. To what extent do the branch managers have a say in promotions within the branch?

(Scale 1 to 7, as described below)

3. To what extent do the branch managers have a say in the hours the branch is open?

(Scale 1 to 7, as described below)

4. To what extent do the branch managers have a say in changing the process for selling a new investment product?

(Scale 1 to 7, as described below)

Scale: 1. Branch decides, senior management has no say.

2. Branch decides, senior management provides advice but no more.

3. Branch decides, senior management influences the decision.

4. Both senior management and branch must approve the decision.

5. Senior management decides, the branch influences the decision.

6. Senior management decides, the branch provides advice but no more.

7. Senior management decides, the branch has no say.

Note: DELEGATE is reverse-coded so that a higher score implies more delegation.

INCENTIVE is the response to the following WFIC survey question to the office of the senior executive of the retail bank:

For the typical branch manager, what percentage of annual pay is bonus or variable?

(Scale: 1 = 0 percent, 2 = 1-6 percent, 3 = 7-10 percent, 4 = 11-15 percent, 5 = 16-20 percent, 6 = 21-25 percent, 7 = greater than 25 percent.)

Firm's Operating Environment

GROWTH:

Growth in insured deposits from June 1994 through June 1995, computed using the FDIC call reports. (12)

STDROA:

The standard deviation of the annual net income (as a percentage of assets to control for heteroskedasticity) from 1990-1994, computed using the FDIC call reports.

INNOV:

The response to the following WFIC survey question to the office of the senior executive of the retail bank:

Please indicate the extent to which the following statement describes your retail bank: We are innovative in the way we design and deliver products/services.

(Scale: 1 = Strongly Disagree, 2 = Disagree, 3 = Neither, 4 = Agree, 5 = Strongly Agree)

SIZE:

Log of the bank's insured deposits of the bank (in $000), computed using the FDIC call reports.

Exogenous Variable for the Delegation Equation

ACQUIRE:

Number of mergers the bank completed between June 1994 and June 1995, collected from the FDIC call reports. To capture merger magnitude, I follow the FDIC categorization and code an acquisition increasing the total assets by < 25 percent as 1, > 25 percent as 2. I then sum the codes for all the bank's acquisitions in this period.

Exogenous Variables for the Incentive Compensation Equation

EDUCATE is the response to the following WFIC survey question to the branch manager of the head branch of the retail bank:

What best describes your level of education?

(Scale: 1 = High School, 2 = Some College, 3 = College Degree, 4 = Some Graduate, 5 = Graduate Degree).

EXPERIENCE is the response to the following WFIC survey question to the branch manager of the head branch of the retail bank:

Have you ever worked for other banks?

(Scale: 1 = Yes, 0 = No).

TABLE 1
Descriptive Statistics for Measures of Branch Manager Delegation,
Incentive Compensation, and the Firm's Operating Environment
in the Retail Banking Industry

                                                            Standard
Variable                                   Mean    Median   Deviation

DELEGATE components:
1. Branch manager discretion over tell-
   ers hired to work in the branch.        2.61     2         1.57
2. Branch manager discretion over pro-
   motions within the branch.              3.53     4         1.18
3. Branch manager discretion over
   hours the branch is open.               4.80     5         1.08
4. Branch manager discretion over
   changing the process for selling a
   new investment product.                 5.94     6         1.00

INCENTIVE                                  2.97     3         1.34

GROWTH                                     4.83%   3.20%     21.29%

STDROA                                     0.32    0.26       0.28

INNOV                                      3.37    4          1.00

SIZE                                       14.17   13.76      1.33

ACQUIRE                                    0.49     0         0.80

EDUCATE                                    2.89     3         1.09

EXPERIENCE                                 0.44     0         0.50

Variable                                   Minimum    Maximum

DELEGATE components:
1. Branch manager discretion over tell-
   ers hired to work in the branch.          1          7
2. Branch manager discretion over pro-
   motions within the branch.                1          7
3. Branch manager discretion over
   hours the branch is open.                 1          7
4. Branch manager discretion over
   changing the process for selling a
   new investment product.                   3          7

INCENTIVE                                    1          7

GROWTH                                     -91.5%      81.4%

STDROA                                       0.02       2.04

INNOV                                        1          5

SIZE                                        10.87      17.86

ACQUIRE                                      0          3

EDUCATE                                      1          5

EXPERIENCE                                   0          1

DELEGATE is the extent of authority delegated to the branch
manager. INCENTIVE is the amount of incentive compensation.
GROWTH is the annual growth in insured deposits. STDROA is
the standard deviation of ROA. INNOV is the firm's strategic
emphasis on innovation. SIZE is the log of insured deposits.
ACQUIRE is the number of acquisitions the bank makes. EDUCATE
is the level of branch manager education. EXPERIENCE is the branch
manager's experience with other banks. Higher scores for
DELEGATE, INCENTIVE, INNOV, EDUCATE, and EXPERIENCE indicate
more delegation, stronger incentives, more strategic emphasis
on innovation, higher education, and more experience,
respectively. The Appendix provides detailed descriptions
of the measures. The number of banks is 100.
TABLE 2
Pearson Correlations among Measures of Branch Manager Delegation,
Incentive Compensation, and the Firm's Operating Environment
in the Retail Banking Industry

              DELEGATE   INCENTIVE   GROWTH   STDROA      INNOV

INCENTIVE     0.19 *
GROWTH        0.18 *     -0.06
STDROA        0.19 *      0.12       -0.11
INNOV         0.27 ***    0.09        0.13    -0.03
SIZE          0.16        0.25 ***    0.02     0.21 **     0.08
ACQUIRE      -0.17 *     -0.15        0.16    -0.05        0.24 ***
EDUCATE       0.03        0.36 ***    0.01     0.21 ***   -0.04
EXPERIENCE    0.07        0.19 *     -0.07    -0.05       -0.02

              SIZE       ACQUIRE    EDUCATE

INCENTIVE
GROWTH
STDROA
INNOV
SIZE
ACQUIRE       0.12
EDUCATE       0.29 ***     0.09
EXPERIENCE   -0.18 *       0.01      0.11

*, **, *** Denotes two-tailed significance at the 10 percent,
5 percent, and 1 percent levels, respectively. DELEGATE is the
extent of authority delegated to the branch manager. INCENTIVE is
the amount of incentive compensation. GROWTH is the annual growth
in insured deposits. STDROA is the standard deviation of ROA.
INNOV is the firm's strategic emphasis on innovation. SIZE is
the log of insured deposits. ACQUIRE is the number of acquisitions
the bank makes. EDUCATE is the level of branch manager education.
EXPERIENCE is the branch manager's experience with other banks.
Higher scores for DELEGATE, INCENTIVE, INNOV, EDUCATE, and
EXPERIENCE indicate more delegation, stronger incentives,
more strategic emphasis on innovation, higher education, and
more experience, respectively. The Appendix provides detailed
descriptions of the measures. The number of banks is 100.
TABLE 3
Simultaneous Estimation of Delegation and Incentive Compensation
for Branch Managers in the Retail Banking Industry

                               DELEGATE
Dependent
Variable               Predicted      Coefficient
Regressors               Sign        (t-statistic)

INTERCEPT                               -2.249 **
                                       (-2.18)

DELEGATE (a)

INCENTIVE                 (-)            0.105
                                        (0.57)

GROWTH                    (+)            0.009 *
                                        (1.92)

STDROA                    (+)            0.583 *
                                        (1.75)

INNOV                     (+)            0.287 ***
                                        (2.92)

SIZE                      (?)            0.064
                                        (0.76)

ACQUIRE                   (-)           -0.298 **
                                       (-2.23)

EDUCATE

EXPERIENCE

Adjusted [R.sup.2]                       0.16 ***
 (second stage)

Adjusted [R.sup.2]                       0.17 ***
  (first stage)

n                                      100

                               INCENTIVE
Dependent
Variable               Predicted       Coefficient
Regressors               Sign         (t-statistic)

INTERCEPT                                 1.312
                                         (0.74)

DELEGATE (a)             (+)              1.145 **
                                         (2.44)
INCENTIVE

GROWTH                   (-)             -0.012
                                        (-1.62)

STDROA                   (-)             -0.657
                                        (-1.18)

INNOV                    (-)             -0.155
                                        (-0.93)

SIZE                     (?)              0.078
                                         (0.72)

ACQUIRE

EDUCATE                  (+)              0.401 ***
                                         (3.33)

EXPERIENCE               (+)              0.240
                                         (0.87)

Adjusted [R.sup.2]                        0.19 ***
 (second stage)

Adjusted [R.sup.2]                        0.19 ***
  (first stage)

n                                       100

*, **, *** Denotes two-tailed significance at 10 percent, 5 percent,
and 1 percent levels, respectively.

(a) Indicates significant endogeneity at the 10 percent level
(Hausman test).

DELEGATE is the extent of authority delegated to the branch manager.
INCENTIVE is the amount of incentive compensation. GROWTH is
the annual growth in insured deposits. STDROA is the standard
deviation of ROA. INNOV is the firm's strategic emphasis on
innovation. SIZE is the log of insured deposits. ACQUIRE is
the number of acquisitions the bank makes. EDUCATE is the
level of branch manager education. EXPERIENCE is the branch
manager's experience with other banks. Higher scores for DELEGATE,
INCENTIVE, INNOV, EDUCATE, and EXPERIENCE indicate more delegation,
stronger incentives, more strategic emphasis on innovation, higher
education, and more experience, respectively. The Appendix provides
detailed descriptions of the measures.

(1) The relation between incentive compensation and delegation is negative in Equation (1), but positive in Equation (2). Such relations are quite common in simultaneous equation models. For example, in the standard Keynesian macroeconomic model of interest rate and output, interest rate negatively affects output, but output positively affects the interest rate (Barro 1993, 551-553).

(2) Holthausen et al. (1995) also use a five-year period to compute the standard deviation of ROA.

(3) The log of deposits is correlated with log of total assets (commercial plus retail) at 0.98.

(4) Prior research finds that the combined bank reduces small-business lending after an acquisition, and that one cannot attribute these reductions to the acquirer cutting out bad or negative net present value loans of the acquired bank (Berger et al. 1998; Sapienza 2002). Since small-business lending is largely the responsibility of lower-level managers, the post-acquisition lending contraction suggests a reduction in these managers' job authority (Stein 2002).

(5) For example, when Travelers, which owned Salomon Smith Barney, merged with Citibank, Salomon Smith Barney bankers' concerns about their bonuses arose primarily from fears of their loss of autonomy in making investment banking deals (Kahn 1998).

(6) One can argue that large firms are more likely to delegate, but this argument is more valid for large conglomerates with diverse divisions rather than large focused firms. Thus, in this argument, size is a proxy for corporate diversification.

(7) The WFIC survey measures EDUCATE and EXPERIENCE only for the head branch, which can be problematic if the head branch is not representative of other branches. I find evidence from the WFIC survey on two features of the head branch that suggest that the head branch is similar to other branches. First, the head branch is not different in size (measured by the number of tellers) from the average branch at the 10 percent two-tailed level. Second, the head branch manager's incentive compensation is not significantly different from that of the average branch manager at the 10 percent two-tailed level. Furthermore, this study focuses on retail branches, and it is unlikely that retail or small-business customers are very different at the head branch (large commercial customers are not handled at the retail level).

(8) For comparison, Baiman et al. (1995, Tables 2 and 3) indicate that incentive compensation for divisional managers in the sample averages about 19 percent of pay.

(9) Statistical research shows that low explanatory power in first-stage estimation does not necessarily cause second-stage coefficient estimates to have large sampling variance (Nelson and Startz 1990).

(10) Low cross-sectional variation in risk could also explain why the risk-premium effect of incentive compensation is insignificant in the delegation equation.

(11) I conduct the factor analysis with oblique rotation, retaining each factor with an eigenvalue greater than unity. Only one factor with an eigenvalue greater than unity emerges, which I label DELEGATE. All four survey questions load on this factor with weights greater than 0.45, the minimum loading threshold recommended in prior literature (Johnson and Wichern 1992, 433; Ittner and Larcker 1995, 7). To construct the DELEGATE factor score, I standardize each survey item to zero mean and unit variance, and sum these items. Prior research recommends the use of such standardized factor scores over actual factor scores (Grice and Harris 1998; Ittner and Larcker 1995, 7). In any event, the standardized factor score for DELEGATE and the actual factor score from the factor analysis are correlated at 0.98.

(12) FDIC Division of Research and Statistics adjusts growth measures for acquisitions as follows. The computation of the growth measures takes into account the values of the acquired institutions for all prior periods, for those institutions that the bank acquires in a pooling-of-interest merger. In contrast to pooling-of-interest accounting, marking-to-market in purchase accounting leads to discontinuities in asset and liability values of the acquired institution. The resulting growth in the combined institution's assets or liabilities over a period that includes the purchase merger could be due to either real growth or marking-to-market. For purchase-accounting acquisitions, the FDIC data have no adjustments, causing performance measures to be distorted. I cannot use the banks' financial reports to correct the data because these reports are available only at the bank holding company level, and not at the individual bank level--the only source of financial data on the individual banks is the FDIC data. However, the distortion appears to be small in the sample: GROWTH's correlation with ACQUIRE, the number of acquisitions, although positive, is insignificant at the 10 percent two-tailed level in Table 2.

REFERENCES

American Bankers Association (ABA). 1998. Retail Banking: An ABA Survey Report. Washington, D.C.: ABA.

Ashkenas, R., L. DeMonaco, and S. Francis. 1998. Making the deal real: How GE Capital integrates acquisitions. Harvard Business Review (January-February): 165-176.

Baiman, S., D. Larcker, and M. Rajan. 1995. Organizational design for business units. Journal of Accounting Research 33: 205-229.

--, and M. Rajan. 1995. Centralization, delegation, and shared responsibility in the assignment of capital investment decision rights. Journal of Accounting Research 33 (Supplement): 135-164.

Baker, G., M. Gibbs, and B. Holmstrom. 1994. The wage policy of a firm. Quarterly Journal of Economics 109: 921-955.

Bank Administration Institute (BAI). 1998. New Paradigms in Retail Banking: The RDS Report on the Future of Retail Banking Delivery--Executive Summary. Chicago, IL: BAI.

Barro, R. 1993. Macroeconomics. 4th edition. Singapore, PRC: John Wiley & Sons.

Beatty, A., S. Chamberlain, and J. Magliolo. 1995. Managing financial reports of commercial banks: The influence of taxes, regulatory capital, and earnings. Journal of Accounting Research 33:231261.

Berger, A., A. Saunders, J. Scalise, and G. Udell. 1998. The effects of bank mergers and acquisitions on small business lending. Journal of Financial Economics 50: 187-229.

Brickley, J., C. Smith, and J. Zimmerman. 1996. Organizational Architecture: A Managerial Economics Approach. Homewood, IL: The Irwin Series in Economics.

Bushman, R., R. Indjejikian, and A. Smith. 1995. Aggregate performance measures in business unit manager compensation: The role of intrafirm interdependencies. Journal of Accounting Research 33 (Supplement): 101-128.

--, --, and M. Penno. 2000. Private pre-decision information, performance measure congruity and the value of delegation. Contemporary Accounting Research 17: 561-587.

Carey, D. 2000. Lessons from master acquirers: A CEO roundtable on making mergers succeed. Harvard Business Review (May-June): 145-154.

Chevalier, J., and G. Ellison. 1999a. Are some mutual fund managers better than others? Cross-sectional patterns in behavior and performance. The Journal of Finance 54: 875-899.

--, and --. 1999b. Career concerns of mutual fund managers. The Quarterly Journal of Economics 114: 389-433.

Collins, J., D. Shackelford, and J. Wahlen. 1995. Bank differences in the coordination of regulatory capital, earnings, and taxes. Journal of Accounting Research 33: 263-291.

Darnell, D. 1999. How NationsBank learned to manage people and change in the merger integration process. Journal of Retail Banking Services 21 (Summer): 1-6.

Demsetz, H. 1992. Comments by Harold Demsetz. In Main Currents in Contract Economics, edited by L. Werin, and H. Wijkander, 275-281. Oxford, U.K.: Basil Blackwell Press.

Grice, J., and R. Harris. 1998. A comparison of regression and loading weights for the computation of factor scores. Multivariate Behavioral Research 33: 221-247.

Holthausen, R., D. Larcker, and R. Sloan. 1995. Business unit innovation and the structure of executive compensation. Journal of Accounting and Economics 19: 279-314.

Horngren, C., G. Foster, and S. Datar. 2000. Cost Accounting: A Managerial Emphasis. 10th edition. Upper Saddle River, NJ: Prentice Hall.

Ittner, C., and D. Larcker. 1995. Total quality management and the choice of information and reward systems. Journal of Accounting Research 33 (Supplement): 1-34.

--, and --. 1998. Are nonfinancial measures leading indicators of financial performance? An analysis of customer satisfaction. Journal of Accounting Research 36 (Supplement): 1-35.

--, and --. 2001. Assessing empirical research in managerial accounting: A value-based management perspective. Journal of Accounting and Economics 32:349-410.

--, --, and M. Meyer. 2001. Anatomy of a balanced scorecard bonus plan. Working paper, University of Pennsylvania, Philadelphia, PA.

Jensen, M. C. 1983. Organizational theory and methodology. The Accounting Review 58: 319-339.

--, and W. Meckling. 1992. Specific and general knowledge and organizational structure. In Main Currents in Contract Economics, edited by L. Werin, and H. Wijkander, 251-274. Oxford, U.K.: Basil Blackwell Press.

Johnson, R., and D. Wichern. 1992. Applied Multivariate Statistical Analysis. Englewood Cliffs, NJ: Prentice Hall.

Kahn, J. 1998. Puzzling it out at Citicorp; Commercial and investment bankers try working together. New York Times (December 18): C1.

Kaplan, R., and A. Atkinson. 1998. Advanced Management Accounting. 3rd edition. Upper Saddle River, NJ: Prentice Hall.

Kirby, A. 1987. Discussion of: Centralization versus delegation and the value of communication. Journal of Accounting Research 25 (Supplement): 19-21.

Kreps, D. 1990. A Course in Microeconomic Theory. Princeton, NJ: Princeton University Press.

Lambert, R., and D. Larcker. 1987. An analysis of the use of accounting and market measures of performance in executive compensation contracts. Journal of Accounting Research 25: 85-125.

Melumad, N., and S. Reichelstein. 1987. Centralization versus delegation and the value of communication. Journal of Accounting Research 25 (Supplement): 1-18.

--, D. Mookherjee, and S. Reichelstein. 1992. A theory of responsibility centers. Journal of Accounting and Economics 15: 445-484.

Milgrom, P., and J. Roberts. 1992. Economics, Organization, and Management. Englewood Cliffs, NJ: Prentice Hall, Inc.

Nelson, C., and R. Startz. 1990. The distribution of the instrumental variables estimator and its t-ratio when the instrument is a poor one. Journal of Business 63: S125-S140.

Prendergast, C. 2000. What tradeoff of risk and incentives? American Economic Review 90 (2): 421-425.

Sapienza, P. 2002. The effects of banking mergers on loan contracts. The Journal of Finance 57: 329-367.

Stein, J. 2002. Information production and capital allocation: Decentralized vs. hierarchical firms. The Journal of Finance (forthcoming).

Stiglitz, J. 1994. Whither Socialism? Cambridge, MA: The MIT Press.

I thank Stanley Baiman, John Core, Adam Zak Gileski, Pierre Hatay, Clement Har, Chip Hunter, Raffi Indjejikian, Christopher Ittner, William Lanen, David Larcker, Russell Lundholm, DJ Nanda, Canice Prendergast, Madhav Rajan, Sofia Trabz, and especially Ken Gaver (associate editor) and two anonymous referees. Comments from the seminar participants at the University of Michigan, Michigan State University, AAA Management Accounting Conference 1999, and the Big 10 Faculty Consortium are greatly appreciated. I am grateful to the Wharton Financial Institutions Center for providing the data.

Submitted March 2000

Accepted September 2001

Venky Nagar
University of Michigan