Minding the Gap

Theoretical Underpinnings of Pay Dispersion

Tournament theory has long served as the cornerstone of pay dispersion research (Connelly, Tihanyi, Crook, & Gangloff, 2014; Henderson & Fredrickson, 2001; Lazear & Rosen, 1981). Tournament theorists propose that employees compete against one another for high level positions and pay. Those who “win” the tournament are promoted to their firms’ top levels and receive higher pay. Firms are resource-constrained with respect to pay, so compensation policies are essentially a zero-sum proposition; that is, increased pay at one level imposes limits at other levels (Bloom, 1999). For this reason, “tournaments are invariably present in common internal labor markets” (Grund & Westergaard-Nielsen, 2008: 486). Labor economists, therefore, originally developed and used tournament theory to help explain the presence of large differentials in prize structures, such as disproportionately high top manager salaries (Ehrenberg & Bognanno, 1990; Lazear & Rosen, 1981; Nalebuff & Stiglitz, 1983).

In organizational contexts, tournament theory’s main prediction is that employee effort increases with differences in pay between organizational levels. Under this view, high upper echelon level pay is thought to effectively motivate employees at all levels to perform well. This view suggests that employees are inspired by the appeal of high pay at the top levels of their firms and, thus, will expend greater effort and commit themselves to organizational interests and priorities (Becker & Huselid, 1992). This is true not only of a final tournament for the CEO position (e.g., Henderson & Fredrickson, 2001), but also those employees striving to reach the upper echelon level (Cappelli & Cascio, 1991; DeVaro, 2006; Lambert, Larcker, & Weigelt, 1993). In support, scholars have shown that top managers and CEOs serve as salient referents to employees; thus, interrank pay differences have important consequences for employee attitudes and behaviors (Wade, O’Reilly, & Pollock, 2006). ¹

Although scholars have sought to understand the implications of widening pay dispersion for firms and the people they employ (Beaumont & Harris, 2003; Wade et al., 2006), this work has failed to reach consensus regarding its effects. One view, primarily using tournament theory arguments, espouses the benefits of large interrank pay differentials, by arguing that providing incentives to rise above others motivates healthy competition, high effort expenditure, and increased performance (Baker, Jensen, & Murphy, 1988; DeVaro, 2006; Ehrenberg & Bognanno, 1990). Some empirical studies have supported this view, finding a positive link between pay dispersion and firm performance (Heyman, 2005; Lallemand et al., 2004; Main et al., 1993).

A second body of research, that incorporates equity theory arguments, proposes the opposite. The assumptions underlying this perspective suggest that because employees look to pay to gauge the fairness of their rewards, pay imbalances motivate feelings of inequity, injustice, and jealousy, which reduces their satisfaction and commitment (see Finkelstein, Hambrick, & Cannella, 2009, for a comprehensive review). Thus, scholars taking this perspective generally assert that pay dispersion tends to reduce motivation, effort, and cooperation (Cowherd & Levine, 1992). Under this view, pay dispersion may also motivate careless risk taking, reduce employee safety (Becker & Huselid, 1992), and ultimately lead to turnover (Bloom & Michel, 2002; Dye, 1984; Gupta et al., 2012). Some empirical evidence supports this perspective, demonstrating a negative relationship between pay dispersion and a number of performance-related outcomes (Bloom, 1999; Fredrickson, Davis-Blake, & Sanders, 2010; Grund & Westergaard-Nielsen, 2008; Siegel & Hambrick, 2005).

As the work above suggests, scholars could draw on one set of prescriptions to explain results that positively link pay dispersion to firm performance or another set to explain negative performance effects. As a result, these two theoretical views are often viewed as competing. Taking a more nuanced perspective, we propose these theoretical assumptions can be reconciled by incorporating temporal effects.

Pay Dispersion and Short-Term Performance

Some research has uncovered evidence in favor of the premise that pay dispersion has positive effects on motivation (Heyman, 2005; Lallemand et al., 2004). Stemming from tournament theory, the main prediction is that actors’ efforts increase with the spread between winning and losing prizes (Knoeber & Thurman, 1994; Lazear & Rosen, 1981). This prediction has enjoyed appreciable empirical support through the years (Gibbs, 1994; McLaughlin, 1988). For example, in a study of 3,510 employers, DeVaro (2006) demonstrated that larger pay spreads positively influenced employee performance. In another study, Knoeber and Thurman (1994) compared the performance predictions of tournament and linear performance evaluation structures, finding that effort and performance were affected more by the difference in pay than by the actual level of pay. An impressive body of empirical management research has uncovered evidence in favor of tournament theory’s basic premise regarding compensation structures (e.g., Cappelli & Cascio, 1991; Conyon, Peck, & Sadler, 2001).

These results notwithstanding, scholars have argued that pay dispersion primarily trains individuals’ efforts on near-term metrics (Marginson & McAulay, 2008). In this way, pay dispersion is argued to motivate a sharp focus on short-term outcomes (Laverty, 1996). For example, as individuals compete with one another for promotion to higher levels of organizations, those doing the promoting often compare candidates on the basis of financial barometers, such as quarterly sales or returns, particularly if they reside in different organizational areas (Brown, Sturman, & Simmering, 2003; Siegel & Hambrick, 2005). In such situations, employees may feel compelled to make choices that offer relatively fast paybacks to rapidly enhance their potential for promotion (Meulbroek, Mitchell, Mulherin, Netter, & Poulsen, 1990; Narayanan, 1985). This could focus their efforts on short-term objectives, designed to increase their personal performance and maximize their potential for promotion (Wahal & McConnell, 2000).

Recent tournament theory research in the marketing literature supports these arguments (Syam, Hess, & Yang, 2011). Given that firms spend more than $26 billion annually on sales contests, marketing scholars have taken a keen interest in how to create optimal designs of rank-order tournaments as a motivational device (Kalra & Shi, 2001; Poujol & Tanner, 2009). Studies show that increasing sales tournament prize money dispersion is highly effective in improving sales peoples’ motivation and, thus, yields greater short-term performance (Garrett & Gopalakrishna, 2010; Kalra & Shi, 2001; Poujol & Tanner, 2009). This application of tournament theory in marketing is consistent with our suggestion that pay dispersion may result in short-term performance benefits for firms. Tournament theorists working in the areas of sport management (Frick, 2003) and law (Price, 2003) have uncovered similar findings with respect to the immediate benefits of prize dispersion in tournaments. Given this evidence, we hypothesize the following:

Pay Dispersion and Long-Term Performance

There are, however, some actions that are effective in the short term but that produce suboptimal results in the long term (Hayes & Abernathy, 1980; Rappaport, 2005). For example, in the marketing research described above, scholars found that high sales tournament reward dispersion can lead to short-term benefits that accrue at the expense of long-term performance as salespeople neglect customer relationship development and fail to meet customer service expectations (Oliver & Anderson, 1994; Poujol & Tanner, 2009). The argument here is that striving for short-term measures can have adverse consequences for the long-term value-generating capability of the firm (Van der Stede, 2000). Laverty (2004) calls this a “temporal trap,” because it exchanges short-term profitability for requisite long-term positioning, growth, and change.

Higher levels of pay dispersion may make employees more vulnerable to these temporal traps because large gaps in pay motivate those individuals to engage in activities that yield early benefits at the expense doing what is right for the firm over the long run (Becker & Huselid, 1992; Bothner, Kang, & Stuart, 2007). There may be, therefore, a disincentive for long-term investment because the potential economic value of investments with payouts beyond the life of a tournament is less observable than the value of investments where managers can realize gains quickly. In this way, some have argued that pay dispersion motivates economic short-termism, wherein a focus on near-term goals precludes a long-term perspective (Laverty, 2004; Marginson & McAulay, 2008).

Within tournament theory, the role of social comparison is helpful in explaining the long-term consequences of pay dispersion. Tournament theorists acknowledge that the existence and awareness of a tournament implies that participants engage in pay comparisons (Lazear & Rosen, 1981). In fact, the effectiveness of promotion tournaments are based on upward comparison between lower-level employees and those who reside at higher levels of the organization. It is this upward pay comparison that is designed to motivate employees to expend more effort to perform better (Lazear & Rosen, 1981). However, individuals also compare their pay to that of others with a view toward making sense of their abilities and gauging the fairness of rewards (Finkelstein et al., 2009). Indeed, tournament theory research has shown that relative assessments of pay are actually even more salient to individuals than are the absolute amounts of pay they receive (Folger & Cropanzano, 1998; Heyman, 2005). Unfavorable comparisons can result in lower morale and reduced motivation, which would likely have adverse effects on the firm’s long-term performance trend (we use the term long-term performance trend to describe the direction of a firm’s performance over time as opposed to a snapshot of performance at any particular point in time).

Consistent with this idea, equity theory research holds that, over time, pay dispersion encourages employees to take corrective action to restore perceptions of fairness. In this way, pay dispersion can reduce cooperation, productivity, commitment, and retention (Bolino & Turnley, 2008; Shaw & Gupta, 2007; Trevor & Wazeter, 2006). In support, equity scholars have found that pay dispersion can motivate competitive behaviors that produce adverse long-term performance effects, such as reckless actions (Becker & Huselid, 1992), political behavior (Dallas, 2011), and uncooperative behavior (Lazear, 1989). Equity theory research also draws attention to a more final consequence, which is increased job exit (Messersmith et al., 2011). Bloom (1999), for instance, notes that by-products of competitive behavior in promotion tournaments can include reduced employee commitment and workforce instability. Thus, equity theory describes how the positive impact of motivating firms’ high performers through interrank pay dispersion could be offset over the long term via lower levels of cooperation and higher levels of turnover (Drago & Garvey, 1998).

Taken together, this research suggests that pay dispersion can lead to having top managers and their employees focusing on short-term results at the expense of long-term goals, social comparisons that reduce morale and motivation, and feelings of inequity among lower-level employees that can trigger a number of mechanisms that adversely affect the firm’s long-term performance trend (Brown et al., 2003; Pfeffer & Langton, 1993). These arguments point to the following hypothesis:

Institutional Investors and Pay Dispersion

A wide range of scholars have shown that institutional investors are influential in affecting firm policies (Connelly, Hoskisson, Tihanyi, & Certo, 2010; David, Hitt, & Gimeno, 2001; Filatotchev & Toms, 2006). This group of investors consists of mutual funds, hedge funds, pension funds, banks, insurance companies, foundations, and endowments that, according to SEC rules, manage more than $100 million. Such investors have considerable clout owing to their centralized voting rights, vast resources, large staffs, and membership in dedicated coordinating bodies, such as Institutional Shareholder Services and the Council of Institutional Investors (Gillan & Starks, 2000). Institutional investors have garnered steadily increasing shares of equities markets, making them a powerful economic force (David et al., 2007). Researchers have demonstrated the extent of their reach with regard to firm activities such as innovation, risk taking, corporate social performance, and compensation policies (Almazan, Hartzell, & Starks, 2005; David, Kochhar, & Levitas, 1998; Hoskisson, Hitt, Johnson, & Grossman, 2002). Building on this work, we focus on the influence of institutional investors on pay dispersion.

Through the years, a number of influential scholars have voiced concerns that equity markets essentially force managers into myopic decision making (Porter, 1992). These authors argue that capital markets operate efficiently only with respect to changing expectations of short-term earnings (Froot, Scharfstein, & Stein, 1993). That is, owing to the rate at which securities turn over, capital markets do not consistently place a premium on long-term value and competitive advantage (Loescher, 1984). Investors, the reasoning goes, have neither the interest nor the knowledge to wait for long-term results, and instead continually push for steadily increasing quarterly results (Matsumoto, 2002). More recently, however, other scholars have advanced this literature by showing that different institutional investors may have dissimilar temporal preferences, and therefore some may be more interested in long-term outcomes than others (Samuel, 2000; Wahal & McConnell, 2000). Acknowledging the varying temporal preferences of institutional investors has gained momentum in the literature as researchers have found this distinction to be consequential to the influence that investors impose on managers (Hoskisson et al., 2002; Tihanyi, Johnson, Hoskisson, & Hitt, 2003). This is likely to be an important distinction with respect to how investors impact pay dispersion (Hartzell & Starks, 2003).

To investigate this influence, we categorize institutional investors according to their preferences (Zheng, 2010). The classification system that is most germane is Porter’s (1992) description of shareholders based on trading behavior. In this system, “dedicated” investors acquire concentrated positions in a small number of firms and hold their shares over time (Bushee, 2004). “Transient” investors, on the other hand, acquire stakes in a wider range of firms and generate substantial turnover of shares in their portfolios (Bushee, 2001). Transient and dedicated investors represent opposite extremes; other investors, such as indexed or exchange-traded funds, are less likely to make trading decisions based on specific firm actions or policies, and are therefore of less theoretical interest.

Transient Institutional Investors

Transient institutional investors move their holdings in and out of firms frequently, leading some to label them as short-term “traders” rather than “owners” (Hendry, Sanderson, Barker, & Roberts, 2006). As a result, transient institutional investors’ preferences for firm outcomes are largely immediate, which can increase short-term pressures on firms. These investors maintain ownership in a broad array of firms, which diminishes their ability to be keenly aware of the operations of any one of their holdings. In addition, transient institutional investors turn their portfolios over frequently, which restricts their ability to develop expertise with respect to particular firms (Loescher, 1984). Furthermore, transient institutional investors are typically beholden to their own constituents that are demanding of consistent positive quarterly returns (Koh, 2007; Marginson & McAulay, 2008). As a consequence, transient institutional investors tend to rely on financial controls and short-term financial barometers of performance. This short-term focus suggests that transient investors are not likely to be concerned about the strategic implications of pay dispersion, which require time to take hold, but instead favor compensation structures that yield near-term benefits.

Another argument regarding the influence of transient investors on pay dispersion surrounds the notion of risk propensity. Pay dispersion could be described as a risky strategy. There are potential rewards, because high top manager pay may have immediate benefits insofar as it assists in attracting well-known managerial talent, which could have signaling value for investors (Higgins & Gulati, 2006). However, there is also downside risk because of its potential implications for the broader organization (Bloom & Michel, 2002). Tolerance for this kind of risk is high for transient institutional investors. These investors maintain broad portfolios with high turnover, so they are less concerned about the risk incurred by any one firm in their portfolio because there is a high probability they will not even hold the firm in the periods ahead (Bushee, 2004; Gompers & Metrick, 2001). Thus, for transient investors, compensation policy risk is mitigated by their diversified and fast-moving holdings.

Based on these ideas, we suggest that higher levels of transient institutional ownership likely result in pressures for short-term, risky strategies, including higher levels of top manager pay and, thus, greater pay dispersion. Some have described owners having influence on firm policies through the mechanisms of exit, voice, and loyalty (Gillan & Starks, 2000; Kochar & David, 1996). In the case of transient investors, it may be that their influence on pay dispersion is less a result of voice or loyalty than of managerial reaction to pressures associated with the threat of exit (Parrino, Sias, & Starks, 2003). Recent work by Dikoli, Kulp, and Sedatole (2009) yields preliminary support for this point of view. These authors hypothesize and find that top manager compensation policies “will reflect firm reactions to the investor base and, in particular, to the presence of high levels of transient institutional investors” (Dikoli et al., 2009: 740). We agree and suggest that firms are both aware of the composition of their investor base and that they make compensation policy choices in accordance with transient-induced short-term pressures. These arguments point to the following hypothesis:

Dedicated Institutional Investors

Dedicated institutional investors, on the other hand, bring considerable resources to their investments, and these resources are concentrated on a small number of firms in a few industries (Bushee, 1998). Accordingly, dedicated institutional investors tend to become familiar with the top managers of firms they own and are often involved in the actions and policies those managers implement (Koh, 2007). This makes dedicated investors more tolerant of firm actions, such as an acquisition or establishment of a subsidiary, that might have negative short-term economic implications but provide long-term competitive value. In the same manner, it makes them less tolerant of activity that could harm the firm’s long-run prospects. Dedicated institutional investors are able to rely on strategic controls when evaluating firm activity, suggesting they are in a position to understand the firm’s actions and appropriately value its policies, including its compensation structures (Chen, Harford, & Li, 2007).

Also, dedicated institutional investors maintain investments over time with little turnover. The potentially negative effects of pay dispersion on organizational outcomes, such as job satisfaction and organizational commitment, manifest themselves over time and are not always readily observable (Winter-Ebmer & Zweimuller, 1999). Given dedicated investors’ long investment horizons, they are likely to be sensitive to such organizational practices that could have detrimental effects on the firm’s trend in long-term performance. Rather than managers simply reacting to the composition of their investor base and the threat of investor exit, dedicated institutional investors may take an active role in influencing a firm’s compensation policies (Dikoli et al., 2009). Consistent with this notion, Hartzell and Starks (2003) show that institutional investors that exercise voice through shareholder proposals and sitting on the board of firms they own do affect compensation practices. Firms with high levels of dedicated ownership, therefore, may be subject to the influence of these long-term investors and adopt lower levels of pay dispersion in accordance with their preferences.

Last, risk propensity arguments are pertinent to this class of investors as well. The level of risk for these owners is even higher than most. Specifically, dedicated investors’ ownership portfolios are highly concentrated in a small number of firms, making them more susceptible to risk factors in any given firm (David et al., 1998). As a consequence, they are vulnerable to organizational practices that hold the potential for introducing volatility. This suggests that top managers in firms with high levels of dedicated ownership might consider it difficult to institute actions that could have negative long-term consequences, such as increasing TMT compensation at the expense of employees. Given these arguments, we propose that dedicated owners are likely to understand the potentially negative long-term implications of a wide gap in pay and discourage the existence of high levels of pay dispersion in firms they own. Therefore, we hypothesize the following:

Sample and Measures

The sample for our study included all publicly traded firms listed in the S&P 1500 composite index during the years 1996 to 2006 that report labor expenses, plus actively traded firms that were once part of the S&P 1500 but have since been removed. The S&P 1500 index combines firms from the S&P 500 large-cap, S&P 400 mid-cap, and S&P 600 small-cap, which provides a comprehensive range of firms and industries for testing the antecedents and consequences of pay dispersion. Standard & Poor’s has added a few other large publicly traded firms to their Execucomp and Compustat databases by client request, and we included these in our sample as well. To identify changes over time we collected data covering an 11-year window, terminating in 2006, which allowed us to calculate a subsequent 5-year performance trend. We eliminated firms with poor reporting and controls by imposing a restriction that they have complete financial data. ² This yielded a sample of 2,410 observations when testing the short-term performance consequences of pay dispersion. The sample was reduced to 1,863 for analysis of the long-term performance trend over 5 years because some firms dropped out of our sample owing to factors such as acquisition or delisting.

We drew the data for this study from four main archival sources. We collected top manager compensation data from the Execucomp database, which provides information about salary, bonuses, and options for CEOs and other top managers. We obtained information about firm-level characteristics and industry-level trends from the Compustat Fundamentals Annual database and product and geographic diversification data from the Compustat Segments database. Last, we gathered data on the firm’s ownership structure from the Thomson Reuters database on institutional common stock holdings and transactions, which is derived from filings of Securities and Exchange Commission (SEC) form 13F.

Control variables

We controlled for several factors that could also influence our dependent variables. First, because prior firm performance is often a strong indicator of future performance and generally associates with firm resources and pay, we controlled for prior year performance as return on assets (ROA) one year prior to the focal year of analysis. Because ROA is the ratio of net income to total assets, it also serves as a control for organization size (including the natural log of assets as a control yielded results consistent with the models presented here). Furthermore, firms with higher levels of product and international diversification are more complex and top managers generally have greater functional and geographic experience than those in less diversified firms, which may allow them to command higher salaries (Carpenter & Sanders, 2004). Thus, we controlled for product diversification and international diversification with inverse Herfindahl indices that increase with higher levels of diversification. We calculated this index for product and geographic diversification by taking the inverse of the sum of squares of the proportion of total sales from each industry or geographic segment in which the firm operates.

We also controlled for three industry level factors that could influence firm performance and pay dispersion: munificence, dynamism, and complexity. We followed the recommendations of Boyd (1995) for these measures. Munificence refers to the capacity of an industry to support sustained growth. We calculated munificence as the regression of industry sales over time divided by mean industry sales, using a 5-year window with the focal year as the last year in the series (Bergh, 1998). Dynamism captures the level of instability or turbulence present in an industry. We calculated dynamism as the standard error of the prior regression divided by mean industry sales (Lepak, Takeuchi, & Snell, 2003). Complexity reflects the inequalities amongst competitors with monopoly or oligopolistic market structures representing less complex competitive environments than those industries closer to a perfect competition framework. We measured complexity with a Herfindahl index, calculated as the sum of squares of market shares of all firms in each four-digit SIC category (Fischer & Pollock, 2004).

We also included a set of year dummy variables to control for temporal effects. Finally, in our models testing the impact of pay dispersion on short- and long-term performance we controlled for dedicated and transient ownership, and in our models predicting pay dispersion we controlled for prior year pay dispersion. Table 1 shows the descriptive statistics for all the variables used in this study.

OPEN IN VIEWER

Table 1

Descriptive Statistics

	M	SD	1	2	3	4	5	6	7	8
1. Pay dispersion	49.45	69.07
2. Return on assets a	3.60	6.76	.10
3. Product diversification	0.23	0.28	.12	.09
4. International diversification	0.10	0.20	.18	.05	.63
5. Complexity	0.14	0.15	−.00	.12	.35	.24
6. Dynamism	0.03	0.03	−.08	.11	.10	.07	.36
7. Munificence	0.09	0.08	−.01	.07	−.02	−.02	.07	.35
8. Dedicated institutional ownership a	6.13	8.57	.07	.06	.10	.10	.06	−.03	−.02
9. Transient institutional ownership a	5.97	8.21	.01	.09	.10	−.00	.06	.08	−.00	.17

Note: N = 2,410. Correlations greater than .04 are significant at p < .05, and correlations greater than .05 are significant at p < .01.

a

Expressed as a percentage.

Analysis

Our data are longitudinal in nature with repeated measures for each firm across multiple years. Using ordinary least squares (OLS) regression increases the likelihood of heteroscedastic error terms and autocorrelation, so panel data of this nature are frequently estimated using fixed effects or random effects models (Devers, McNamara, Wiseman, & Arrfelt, 2008). We conducted a Hausman (1978) test, which indicated the appropriate choice was a fixed effects model (χ² = 880.54, p < .001). Fixed effects models provide within-firm regression estimation and a conservative test of our hypotheses (Sanders, 2001). We used robust standard errors in all analyses (though results without using robust standard errors were consistent with those presented here).

We first tested the influence of pay dispersion on short-term performance in the focal year (Time t) while controlling for prior performance (Time t-1). Table 2 shows the tests of Hypothesis 1, which predicts that higher pay dispersion is positively associated with short-term performance. In Model 1, the only significant control variable was transient ownership, indicating that higher transient ownership was positively associated with short-term performance. In Model 2, the coefficient for pay dispersion was positive and significant (p < .01), supporting Hypothesis 1.

OPEN IN VIEWER

Table 2

Fixed Effects Panel Regression of Short-Term Firm Performance

Variable	Model 1	Model 2
Constant	1.45	2.05*
	(0.95)	(1.00)
Control variables
Prior year performance	0.12	0.11
	(0.10)	(0.10)
Product diversification	0.00	0.00
	(0.01)	(0.01)
Geographic diversification	−0.05	−0.05
	(0.04)	(0.04)
Complexity	−0.01	−0.01
	(0.03)	(0.03)
Dynamism	−0.02	−0.01
	(0.06)	(0.06)
Munificence	0.02	0.01
	(0.02)	(0.02)
Dedicated ownership	−0.00	−0.00
	(0.03)	(0.07)
Transient ownership	0.08**	0.07*
	(0.03)	(0.03)
Year (effects coded)	Included	Included
Independent variable
Pay dispersion		0.01**
		(0.00)

Note: N = 2,410. Robust standard errors are in parentheses.

*

p < .05. **p < .01. One-tailed tests for hypothesized variables, two-tailed tests for control variables.

We then tested the influence of pay dispersion on the long-term performance trend (Time t to Time t+4), again controlling for prior performance (Time t-1). Models 3 and 4 in Table 3 show our tests of Hypothesis 2. As the results of the control model reveal, prior performance was a negative predictor of long-term performance. Next, as Model 4 shows, the influence of pay dispersion on the long-term performance trend was negative and significant (p < .01), supporting Hypothesis 2.

OPEN IN VIEWER

Table 3

Fixed Effects Panel Regression of the Long-Term Firm Performance Trend

Variable	Model 3	Model 4
Constant	0.06	−0.203
	(0.69)	(0.72)
Control variables
Prior year performance	−0.08**	−0.08**
	(0.03)	(0.03)
Product diversification	−0.00	−0.00
	(0.01)	(0.01)
Geographic diversification	0.02	0.02
	(0.02)	(0.02)
Complexity	−0.03	−0.03
	(0.02)	(0.02)
Dynamism	−0.04	−0.04
	(0.06)	(0.06)
Munificence	−0.02	−0.02
	(0.01)	(0.01)
Dedicated ownership	−0.03	−0.02
	(0.02)	(0.02)
Transient ownership	−0.01	−0.01
	(0.01)	(0.01)
Year (effects coded)	Included	Included
Independent variable
Pay dispersion		−0.01**
		(0.00)

Note: N = 1,863. Robust standard errors are in parentheses.

**

p < .01. One-tailed tests for hypothesized variables, two-tailed tests for control variables.

Last, we tested the influence of dedicated and transient ownership as antecedents (at Time t-1) of pay dispersion in the focal year (Time t). Models 5 and 6 in Table 4 show our tests of Hypotheses 3 and 4. Tests of the control variables revealed that product diversification levels and prior firm performance positively influence pay dispersion, suggesting that top managers appear to be disproportionately rewarded when firms are diverse and have performed well (e.g., Devers et al., 2007). As shown in Model 6, the coefficient for transient investors was positive and significant (p < .05), supporting Hypothesis 3. The coefficient for dedicated investors was negative and significant (p < .05), supporting Hypothesis 4. Entering the transient and dedicated institutional investor variables separately does not change these results.

OPEN IN VIEWER

Table 4

Fixed Effects Panel Regression of Pay Dispersion

Variable	Model 5	Model 6
Constant	−2962.62	−2702.03*
	(1096.32)	(1093.14)
Control variables
Prior pay dispersion	−0.04	−0.04
	(0.06)	(0.05)
Prior year performance	59.21*	57.31*
	(26.91)	(26.82)
Product diversification	−15.83*	−17.33*
	(7.86)	(7.91)
Geographic diversification	−8.41	−9.10
	(14.79)	(14.83)
Complexity	61.89	60.54
	(47.04)	(47.30)
Dynamism	−14.73	−22.11
	(50.28)	(51.67)
Munificence	16.85	19.76
	(14.20)	(14.40)
Year (effects coded)	Included	Included
Independent variables
Transient ownership		25.79*
		(12.65)
Dedicated ownership		−31.88*
		(17.05)

Note: N = 2,292. Robust standard errors are in parentheses.

*

p < .05. One-tailed tests for hypothesized variables, two-tailed tests for control variables.

Supplemental Analyses

The possibility exists that our pay dispersion variable was endogenous. Therefore, we conducted an additional set of analyses to determine whether potential endogeneity was influencing our findings. In this set of analyses we reestimated our models using the xtabond2 procedure in STATA, which utilizes the generalized method of moments (GMM) model also known as system GMM. The xtabond2 procedure is designed for panels that may contain fixed effects and heteroscedastic and correlated errors within units, and employs first differencing, which instruments variables with suitable lags of their own first differences, to eliminate these issues and potential sources of omitted variable bias (please see Arellano & Bover, 1995; Blundell & Bond, 1998; Roodman, 2009). Furthermore, and importantly, xtabond2 also allows the ability to specify variables as endogenous to examine whether potential endogeneity is influencing findings.

To run our analyses of the supplemental tests of Hypotheses 1 and 2, we used the two-stage GMM estimator procedure in xtabond2 (that employs robust standard errors). In doing so, we specified pay dispersion as an endogenous variable with a complete lag structure. The results of these analyses were entirely consistent with those of the first two sets reported in the manuscript in Tables 2 and 3. This, and tests of overidentifying restrictions (Sargan & Hansen J statistics), demonstrated that our instruments were exogenous, suggesting that potential endogeneity was not biasing our results. In addition to the tests above, we also reestimated our models using short-term (annual) and long-term (trend over 5 years) measures of return on sales and found results consistent with those reported above.

Pay Dispersion Research

Perhaps most important is our finding that intrafirm, interrank pay dispersion has significant implications for organizational performance. Although empirical findings of studies examining the effects of pay dispersion within organizations have been conflicting, less research specifically examines interrank pay dispersion between the highest and lowest levels of the organization (Eriksson, 1999). Those that do are usually limited to a subset of the organization: top, middle or bottom (Brown et al., 2003; Siegel & Hambrick, 2005). Few studies have examined top-to-bottom, vertical pay dispersion such as that which companies must report to be in compliance with the Dodd-Frank Act (e.g., Eriksson, 1999). As a result, the literature on pay dispersion has reflected “a certain fragmentariness or case-study quality across the studies, which examine specific, usually narrow labor markets” (Grund & Westergaard-Nielsen, 2008: 487). Our study helps smooth these spiky empirical results by investigating pay dispersion in an unusually rich and expansive set of firms.

Findings appear to show that pay dispersion has more nuanced implications for organizational performance than currently acknowledged. Scholars often draw on tournament theory assumptions to explain results that positively link pay dispersion to firm performance and equity theory assumptions to explain negative performance effects. In this study we hypothesize and find that, rather than contrasting, both tournament theory and more equity-based arguments can help explain the effects of pay dispersion when temporal differences are considered.

Our study also provides some methodological advances for research on pay dispersion. Empirical research vertical, intrafirm pay dispersion has been decidedly narrow (Takacs-Haynes, 2008). While there have been studies in a number of industries, few have attempted to simultaneously examine firms in multiple industries. We hope, therefore, that the combination of a longitudinal examination of both short- and long-term performance indicators across industries adds an element of comprehensiveness to the empirical body of pay dispersion research.

Future research on pay dispersion might build on our findings to further our understanding of its antecedents and consequences. Beyond the firm-level consequences discussed in our study, there may be other meaningful consequences of widening pay gaps, such as effects on the speed or complexity of a firm’s competitive actions. Similarly, higher levels of pay dispersion might also influence a firm’s relationships with customers and suppliers. We would encourage scholars working in the areas of competitive dynamics and interorganizational relationships to explore these possibilities. Furthermore, scholars might extend our findings by considering other potentially important drivers of pay dispersion. Given our findings regarding competing short- and long-term performance outcomes, there may be other governance factors, such as board characteristics or the market for corporate control, that change the likelihood of firms adopting higher levels of pay dispersion. We expect social network research might also be informative in this regard and could help explain how higher levels of pay dispersion diffuse among a network of organizations.

Tournament Theory Research

The results shown here also extend the literature on tournament theory. Tournament theory has gained tremendous momentum in management research as a means to explain variation in compensation levels and practices (Connelly et al., 2014; Gomez-Mejia, Trevino, & Mixon, 2009; Messersmith et al., 2011). Its most prominent prediction is that wider differences in pay motivate more effort and greater performance (DeVaro, 2006). However, if, as our results suggest, pay dispersion yields short-term benefits with negative long-term repercussions, a natural question is this: To what extent do those responsible for controlling pay dispersion allow, or even facilitate, its growth? In this study, we address this question by developing and testing a theoretical explanation of ownership antecedents of pay dispersion. We believe that by beginning to shed light on the drivers of pay dispersion between top-level managers and non-top-level managers, our study adds an element of comprehensiveness to this important and influential body of work.

Tournament theorists might extend our work by exploring variations of our theory to uncover boundary conditions, or potential moderators, of our hypothesized relationships. For instance, an underlying assumption of tournament theory is that employees are strongly motivated by compensation, but this may be truer in some countries than in others (Takacs-Haynes, 2008). Therefore, empirical examination of how cultural factors change the effectiveness of compensation tournaments could yield interesting results (Oxelheim & Randoy, 2005). Also, a moderating influence on actor effort in tournaments could be the number (McLaughlin, 1988) or type (Nippa, 2010) of people competing in the tournament. Thus, our findings regarding the effects of pay dispersion could be moderated by the number of levels in the organization, the number of people competing at each level, or the culture of the organization. Furthermore, our findings regarding different owner preferences for tournament-based compensation mechanisms suggest there may be room for integrating agency theory with tournament theory. In particular, multiple agency theory (Arthurs, Hoskisson, & Busenitz, 2008) may help describe how managers address the competing interests of principals.

Implications for Practice

Though there are exceptions, we expect that top managers are generally interested in increasing their pay. However, our study draws attention to two cautions those managers should observe. First, increasing pay dispersion could have important negative long-term ramifications both for their firms and the people they employ, thereby placing their financial, human, and social capital at risk (Heyman, 2005; Winter-Ebmer & Zweimuller, 1999). Second, as our study shows, pay dispersion between top manager and other employees is dependent, in part, on the ownership structure of their firms. Therefore, managers interested in following the trend of those who have expanded pay dispersion in their firms would do well to consider their existing governance structures, what kind of owners they wish to attract, and actions they might take to woo those potential shareholders (Bushee, 2004).

Furthermore, anecdotal evidence suggests that increasing pay dispersion could have important policy implications. For example, our findings about institutional investor influence on pay dispersion may motivate regulatory bodies, such as oversight committees, unions, or government agencies, to more powerfully focus their energies on methods designed to force influential investors to support firm policies and structures that reward long-term performance. Pay dispersion may trigger actions that have important consequences for human resource managers by affecting recruitment, selection, and retention of both top managers and other employees. We suggest that developing a more complete understanding of how pay dispersion influences human resource policies and practices, TMT recruitment and retention, and other stakeholders offers interesting and potentially fruitful research opportunities.