Serving Others at the Expense of Self: The Relationship Between Nonprofit CEO Compensation and Performance in Trade and Professional Associations

This Article is brought to you for free and open access by the School of Public Service at ODU Digital Commons. It has been accepted for inclusion in School of Public Service Faculty Publications by an authorized administrator of ODU Digital Commons. For more information, please contact digitalcommons@odu.edu. Repository Citation Saitgalina, Marina; Bennett, Andrew A.; Reina, Christopher S.; and Coombs, Joseph E., "Serving Others at the Expense of Self: The Relationship Between Nonprofit CEO Compensation and Performance in Trade and Professional Associations" (2018). School of Public Service Faculty Publications. 35. https://digitalcommons.odu.edu/publicservice_pubs/35

egregious.As such, some states are considering limits on nonprofit CEO salaries (Meiksins, 2013).
Arguably, nonprofit CEO compensation is subject to different rules than corporate CEO compensation due to a nondistribution constraint.As one of the tax-exemption requirements, the nondistribution constraint legally bars nonprofits from distributing any excess earnings to those in control of the organization (IRS, 2017).This highlights that the predictors of nonprofit CEO compensation may be different than those for for-profit CEO compensation, although common factors such as market size, organizational scope, number of employees, and annual budget all likely still play a role in determining nonprofit CEO compensation levels (Grasse, Davis, & Ihrke, 2014;Hallock, 2002;Frumkin & Keating, 2010;Newton, 2015).Interestingly, some nonprofit hospitals in the United States tie their CEO compensations to turnover and financial performance indicators (Brickley & Van Horn, 2002), and research on CEO compensation in art and education organizations has demonstrated a positive relationship on organization's performance (Carroll, Hughes, & Luksetich, 2005).
The ongoing conversations about CEO pay and organizational performance in nonprofit organizations are also a result of the pressure to become more effective and efficient, where both lower relative pay and higher relative performance foster a reputation of a good steward of public money.This strategy can also be used intentionally to boost an organization's public profile, improve internal employee motivation, or as a sign of dedication to the mission and the sector.And considering that nonprofits are limited in the number of instruments to affect performance compared with their for-profit counterparts (i.e., nonprofits cannot pay dividends), the issue of how CEO pay affects organizational performance may be especially salient for a broad array of stakeholders.In addition, given the existence of current high-profile examples of corporate CEOs reducing their compensation to increase employee salaries (Isidore, 2015) or to boost firm performance (MacMillan, 2013), we question whether this also happens in the nonprofit sector.Regardless of the distribution constraint, there seems to be a link between CEO compensation and organizational performance in nonprofit organizations.Furthermore, considering the diversity of the nonprofit sector and how it operates, one can expect a certain level of variation in nonprofit compensation incentives (Newton, 2015).
Reflecting on this, we were intrigued by the question as to whether some CEOs knowingly (and on purpose) earn less than their peers.If so, does this have an impact on organizational performance?While there is a plethora of factors that predict nonprofit CEO compensation, what has been explored less is the impact that nonprofit CEO compensation has on organizational performance.Thus, this paper investigates the impact of CEO compensation on organizational performance in a subsector of mutual benefit/membership nonprofits.This type of nonprofit organization consists of trade and business leagues, professional associations, and membership societies that serve the interests of a specific industry or profession and provide benefits to dues-paying members (Reilly, Hull, & Braig Allen, 2003).Theoretically, mutual benefit organizations are organized for the self-interest of their members.Some describe this type of nonprofit as bordering nonprofit and for-profit sectors, in which commercial and nonprofit activities are combined to meet client or member demands (Fine, Ropa, & Jay, 2008).
However, these organizations benefit society by promoting good citizenship in industry and commerce.It is represented through activities such as the establishment of professional standards and performance, certification and licensing, ethical guidance and codes of conduct, professional service values advocacy, and research and knowledge creation, all of which will result in higher public safety outcomes.The role of serving the public provides a rationale for their nonprofit exemptions.Therefore, the broad benefits to the public are found in the professional membership association's mission to strengthen the profession, and the narrow functions they serve are reflected in membership benefits.
Most research on professional associations focused on the membership side of organizational activities, such as joining, volunteering, and retention (Gazley, 2013;Bauman, 2008), whereas a lot less is known about management and governance of these organizations.Haynes and Gazley (2011) noted that the lack of empirical research in the context of professional membership associations have set back the knowledge expansion about this subfield of nonprofit organizations, where most research is focused around traditional charitable types of nonprofits.Thus, studying professional membership associations, which reside at the intersection of the sectors, provides a fruitful and unexplored area of scholarly research.This research expands knowledge about organizations that are not a usual part of the core of research in the nonprofit sector and contributes to the literature in three primary ways.First, we focus on relative CEO compensation, using a framework drawing from social comparison and servant leadership rather than agency theory, a comparison still under discussion in the nonprofit sector (e.g., Van Puyveldeet, Caers, Du Bois, & Jegers, 2016).Second, we examine a sample of CEOs who have access to peer compensation data, a feature of our study that has not been available in previous studies.Third, we utilize both financial and nonfinancial performance measures to address multiple outcomes valued by nonprofits.We begin with a literature review of CEO compensation and performance and then draw from socal comparison and servant leadership literatures to develop and test six hypotheses concerning the relationship between CEO relative compensation and nonprofit performance.

Literature Review
Considering complexities accompanying typical nonprofits, such as nondistribution constraint and hard-to-measure performance indicators, one would expect a limited effect of compensation on performance (Oster, 1998).From a number of studies looking at the effects of compensation on performance, Carroll, Hughes, and Luksetich (2005) found that, while performance increases as executive compensation increases in larger organizations, the inverse is true in smaller organizations.Similarly, Frumkin and Keating (2010) found that nonprofit CEOs compensation is comparable with other similar-sized organizations, implying that executive salaries depend upon the organization's size and suggesting that nondistribution constraint is not strictly enforced.
Given the varying conclusions about the pay-performance relationship, Grasse et al. (2014) inferred that different types of nonprofits may be sensitive to different measures of performance.On the one hand, Baber, Daniel, and Roberts (2002) noted that, despite the fact that performance oftentimes is difficult to quantify, they found a positive correlation between program spending and managers' compensation in a sample of charitable organizations in Maryland, assuming that managers are rewarded for increases in direct program spending.On the other hand, Newton (2015) found a significantly negative relationship between CEO-toemployee relative pay and multiple measures of nonprofit performance (i.e., program spending ratio) in a sample of large charitable nonprofits.Overall, we see a sector that may be rewarding one measure of performance (i.e., funding directed toward programs) but may be ignoring other indicators.

Nonprofit Performance
Measuring organizational performance can be more difficult with nonprofit research, as outcomes such as return on assets (ROA), return on equity (ROE), or other financial measures are not readily available or applicable.One major concern with assessing financial performance is that different measures are used in research, making it difficult to compare financial performance outcomes across studies (see Ritchie & Kolodinsky, 2003, for review).An additional concern regarding nonprofit performance is that performance is often reported by board members and therefore does not represent an objective measure comparable to objective measures commonly found in strategic management research (Brown, 2005).This problem becomes even more pronounced when outcomes used by the board or CEO are unique to the organization because the relationship to financial performance is unclear (Zimmermann & Stevens, 2006).Even if objective performance measures were available and comparable across organizations, Kaplan (2001) argues that financial measures "are not sufficient to motivate and evaluate mission accomplishments" (p.353).Thus, it makes sense that organizations that exist to serve the public good must also consider nonfinancial measures of organizational performance.
In order to capture organizational performance beyond typical accounting measures, individuals and institutions track how efficiently charitable organizations use their funds (Charity Navigator, 2012;Kistruck, Qureshi, & Beamish, 2013).These efficiency measures are a comparison of the money used to serve constituents in relation to the administrative expenses used to run the organization.Nonetheless, these efficiency measures do not tell the entire story.Mitchell (2013) found that nonprofit managers considered goal attainment as even more important than financial effectiveness or efficiency.Miller (2002) suggests some relevant measures to assess nonprofit organization goal attainment include volunteer service (i.e., hours or number of unique volunteers), membership growth, or the amount of money donated rather than revenue.Given this past research and the practical knowledge that nonprofit organizations differ in important ways from for-profit entities, our study aims to build upon research in the nonprofit sector by including multiple measures of performance, namely, both financial efficiency as well as a nonfinancial measure of the number of individuals served during the year.We believe this dual approach to measuring organizational performance most effectively incorporates a nonprofit CEO's emphasis on both improving organizational outcomes and benefiting individuals the organization serves.

Theoretical Development
Agency theory, which unravels relationships in which one party determines the work to be done while another party performs that work, has typically been used when examining the relationship between CEO compensation and organizational performance (Eisenhardt, 1989).However, agency theory does not explain well the intricies of the role of the board of directors in monitoring the CEO in the nonprofit setting (Cumberland, Kerrick, D'Mello, & Petrosko, 2015).One limitation of using agency theory to understand nonprofit leadership is that it does not adequately focus on the desire to help a broad array of stakeholders and to benefit the community (Pepper & Gore, 2015).A second limitation is that agency theory does not take into account reasons why nonprofit leaders may sacrifice their own well-being (Dempsey & Sanders, 2010).Olson (2000) proposed that, although there is some support for agency theory in a nonprofit setting, other psychological factors play an important role in understanding these relationships.
We suggest that servant leadership theory and social comparison theory can help shed some light on these limitations, and thus we utilize these two theoretical frameworks to better understand nonprofit CEO compensation and explain why a nonprofit CEO may knowingly work for lower compensation than his/her peers.This contributes to the nonprofit and management literature by responding to the call for alternative theoretical paradigms that explore how people engage in both self-sacrificing and self-serving behaviors as well as how these behaviors have an impact on organizations (Krause, Semadeni, & Cannella, 2014).

Serving Others Before Self
Servant leaders focus on follower empowerment and growth primarily through meeting followers' needs (Liden, Wayne, Zhao, & Henderson, 2008;van Dierendonck, 2010).Servant leaders are unique in that, instead of the typical top-down approach to leadership, which mirrors a pyramid structure (i.e., one leader at the top being served by multiple followers), servant leadership flips the pyramid upside down such that the leader "serves" organizational members (Greenleaf, 1977;van Dierendonck, 2010).Nonprofit and public service leadership requires motivation toward doing good for the community and willingness to mobilize and build social capital to achieve the social mission of the organization (King, 2004;Quarter & Richmond, 2001), which mirrors servant leader motivation to serve others.Such leaders are intrinsically motivated to carry out their work rather than motivated by extrinsic means, and have a strong overlap between the organization's social mission and their personal values and goals (Perry, Hondeghem, & Wise, 2010;Wright, Moynihan, & Pandey, 2012).Given that nonprofit CEOs are motivated by intrinsic factors and by achieving a social mission, we suggest that it is more likely that these CEOs will be less concerned with their pay relative to their peers at other organizations.According to past research, these servant leaders will be concerned with promoting high procedural justice climates in their organizations (Ehrhart, 2004), which means that their compensation should be in line with what the organization can pay according to its unique attributes.In addition, servant leaders sacrifice extrinsic gain for the benefit of others (Conger & Kanungo, 1994;Liden, et al., 2008).

Comparing Others With Self
Social comparison theory states that individuals assess their own abilities and potential relative to others (Festinger, 1954).Comparison with others provides a necessary benchmark that allows people to make judgments about themselves on various attributes, and it also allows one to estimate his or her own likelihood of success or ability to accomplish something, such as learning a new skill or changing careers.The motivation for comparison, however, is not to significantly outperform others but ultimately to achieve relative uniformity with others by being only slightly better (Festinger, 1954).The theory has been extended to explain such phenomena as an increase in individual charitable contributions during a fundraising campaign when the level of donations of others was known (Shang & Croson, 2006), an increase in firm performance at low pay disparity levels (Ridge, Aime, & White, 2015), and the degree of sensitivity between pay and performance within firms (Gartenberg & Wulf, 2017).

Hypotheses
Previous examinations of the relationship between relative CEO compensation and organizational performance have found mixed results (Frumkin & Keating, 2010;Grasse, Davis, & Ihrke, 2014;Carroll, Hughes, & Luksetich, 2005).We believe this may be because these studies typically use a "market worth" variable comprised of multiple inputs such as organizational performance.One potential concern with this methodological approach of creating a "market worth" variable is that a CEO will use a peer group of similar organizations rather than an arbitrary variable created by scholars.Empirical evidence supports this claim, finding that nonprofit CEO compensation is more strongly predicted by CEO compensation of similar-sized organizations than by organizational performance (Frumkin & Keating, 2010).A second potential concern is that researchers have largely focused on the negative aspects of compensation inequality (Devers, Cannella, Reilly, & Yoder, 2007;Pfeffer & Langton, 1993) without discussing possible positive aspects of this inequality.Specifically, while past research has shown that social information like external comparisons can influence leader self-interested behaviors in the for-profit world (Rus, van Knippenberg, & Wisse, 2010), we suggest that this might not occur in a nonprofit environment.In fact, we go one step farther to consider whether the opposite may be true for nonprofits, namely, that the use of social information can lead to unselfish behavior.This unselfish behavior "Communicates the relatively unambiguous message that the leader has a progroup orientation" (van Knippenberg & van Knippenberg, 2005, p. 26).Thus, we consider the possibility that a CEO is willing to accept unequal compensation relative to actual peers to positively influence organizational performance.

Financial and Nonfinancial Performance
Nonprofit performance is a multifaceted concept.The absence of sole focus on the bottom line uniquely positions nonprofits to achieve multiple and complex goals.Thus, their performance can best be captured by a variety of measurements.Although there is no consensus as to what those measures are, conventionally accepted dimensions of nonprofit performance are financial (e.g., revenue diversification, overhead cost, and cost savings) and nonfinancial measures (e.g., number of clients served, number of volunteers participated, and satisfaction levels), where financial and nonfinancial categories are complementary rather than mutually exclusive (Pandey, Kim, & Pandey, 2017).Boateng, Akamavi, and Ndoro (2016) indicated that a more accurate representation of nonprofit performance is achieved by using a combination of factors.In this vein, we are utilizing two measures of nonprofit performance-financial and nonfinancial-to capture the distinct multidmensionality of the nonprofit sector.
According to previous work by Strachan and Myslewski (1997), looking specifically at professional associations, the strategy to determine executive compensation should include consideration of three components: the business environment (e.g., levels of competition and general directions of the market), culture and management philosophy (e.g., external pressures of efficiency and effectiveness), and compensation philosophy (e.g., comparative pay levels across associations).From this standpoint, which aligns with social comparison theory, an organization will try to perform comparatively better to improve organizational status in the competitive market.And, according to servant leadership theory, accepting a lower compensation can be a motivational strategy for a CEO to encourage higher performance in a culture that values its employees by minimizing the pay difference.
Hypothesis1: Lower CEO compensation relative to peer CEOs will be related to higher nonfinancial organizational performance.
Hypothesis2: Lower CEO compensation relative to peer CEOs will be related to higher financial organizational performance.

Organizational Size as a Moderator
We propose that organizational size moderates the relationship between relative CEO compensation and organizational performance.Considering that organizational size greatly impacts the relationship between relative CEO compensation and organizational performance (Brown, 2005), we test four total hypotheses that investigate how organizational size can alter the relationships found in Hypotheses 1 and 2. We test these hypotheses utilizing two distinct measures that account for a nonprofit organization's size (i.e., organization's annual budget and number of full-time employees), and we discuss each of these in greater detail in the methodology section.
We propose that the relationship between nonprofit CEO compensation and organizational performance will depend on organizational size.Smaller organizations must contend with the liabilities of smallness (Aldrich & Auster, 1986), which increases the difficulty of raising financial capital and acquiring critical human and social capital relative to their larger, more established counterparts.For smaller organizations, these resources are difficult to acquire because external stakeholders such as customers, suppliers, employees, and investors tend to prefer interacting with larger firms because their reliability and legitimacy are well established (Hannan & Freeman, 1984;Stinchecombe, 1965;Stuart, 2000).As a result, CEOs of smaller firms must be highly involved with the organization's day-to-day operations as resources are less available to hire additional staff with which to share decision-making responsibilities.CEO actions and decision-making in smaller organizations are therefore more closely associated with performance outcomes than in larger firms with a decentralized decision-making structure.In addition to CEOs being more involved with decision-making in small firms, they also may have fewer resources available to be dedicated to executive compensation that can be used to attract stakeholders.For example, Barnes, Harikumar, and Roth (2006) reported that, for smaller organizations, founder CEOs accept less compensation to avoid harming relationships with external stakeholders.As a result, we would expect smaller organizations to benefit more from lower CEO compensation because CEOs are more involved in day-to-day decision-making, and financial capital not used for compensation may be used to further support the organization's mission (i.e., serving members), which ultimately improves financial efficiency.
Hypothesis3: Annual budget moderates the relationship between relative CEO compensation and organizational performance, such that the negative relationship between CEO relative compensation and organizational nonfinancial performance will be stronger in organizations with a smaller annual budget.
Hypothesis4: Annual budget moderates the relationship between relative CEO compensation and organizational performance, such that the negative relationship between CEO relative compensation and organizational financial performance will be stronger in organizations with a smaller annual budget.
A second reason that organizational size might influence the relationship between relative CEO compensation and performance is that, in smaller organizations, CEOs will likely have a more direct influence on individual employees.In large corporate settings, most members of large organizations rarely come into contact with executives (Mowday & Sutton, 1993), which is likely the exact opposite experience of most employees working in small nonprofit organizations.Thus, the size of the organization has an important effect on employees' daily experiences.For example, we expect that, in smaller organizations, employees have daily contact with the CEO and likely routine contact with the board of directors, which is not likely the case in large nonprofit organizations.For example, as president and CEO of YMCA of Greater Springfield, James O'S. Morton managed staff and oversaw the YMCA's internal operations and programs on a daily basis, often interacting closely with the YMCA staff and leadership team to accomplish these goals (Bridgespan, 2017).As a smaller nonprofit, being involved in the day-to-day operations as well as being responsible for setting the strategic direction all fell under his responsibility.We thus suggest that the number of employees can be used as a proxy to indicate firm size.Due to the tight-knit nature of small organizations, we suggest that employees may be more likely to be aware that their CEO is being paid less than other nonprofit CEOs of similarly sized organizations and argue that, given this knowledge, they will be inspired to work hard, improve financial efficiency, and ultimately serve more stakeholders, which leads to better organizational performance.Therefore, we propose: Hypothesis5: The number of employees moderates the relationship between relative CEO compensation and organizational performance, such that the negative relationship between CEO relative compensation and organizational nonfinancial performance will be stronger in organizations with fewer employees.
Hypothesis6: The number of employees moderates the relationship between relative CEO compensation and organizational performance, such that the negative relationship between CEO relative compensation and organizational financial performance will be stronger in organizations with fewer employees.

Sample
The sample for this study is drawn from the American Society of Association Executives' (ASAE) 2010 Association Executive Compensation Benefits Survey.The ASAE represents over 7,400 trade associations, individual membership societies, and voluntary organizations across the world (ASAE, 2017).Although trade associations can be incorporated under various tax-exempt designations, this paper narrows the focus to "public serving" 501(c)(3) organizations, which are also precluded from distributing any financial surpluses "to those who control the use of organizational assets" (Powell & Steinberg, 2006, p. 1).The 2010 survey included data on 1,262 organizations of which 290 were 501(c)(3) organizations.In these organizations, CEOs are responsible for "governance, change management, staff performance, and the organization's progress toward success" (ASAE, 2017).
Because the focus of this study is to investigate organizational performance when CEOs know how their compensation compares with that of others, we only use data from organizations that used a compensation consultant or compensation benchmark to determine CEO salary.In other words, we only draw from a sample of organizations in which the CEO would have an accurate understanding of their salary relative to other CEOs.Although some research shows that corporate CEO compensation is higher in firms when compensation data or consultants are used (Murphy & Sandino, 2010), nonprofit CEOs may use their influence to be paid less than their market value.This objective measure comparing relative salary is preferred over self-reported comparisons because it is more resistant to social desirable responding (Tomassetti, Dalal, & Kaplan, 2016).After removing cases with missing data, our final sample is 154 organizations.

Dependent Variables
Efficiency and the number of members served by the organization are the two outcomes of interest in our study.Consistent with Kistruck et al. (2013), efficiency is measured as organization expenditures specific to their charitable programs divided by total organization expenditures.Three survey questions were used to construct this variable, of which two were combined to capture organization nonprogram expenditures and one question to gauge total organization expenditures (see Table A1 in Appendix).Efficiency reflects an organizations' ability to limit administrative and fund-raising expenses while maximizing the amount spent directly on programing (Kistruck et al., 2013).
Our second dependent variable, individuals served, reflects the number of individuals served by the organization and thus represents the total number of individuals that benefit from the organization's mission.We measured this using the survey question, "Approximately how many members or individuals are served by your organization (include all membership classes and leave blank, if not applicable)?"As the number of individuals served by the organization is not normally distributed (Kolmogorov-Smirnov and Shapiro-Wilk tests p < .01),we used the natural logarithm of individuals served to normalize the distribution.

Independent Variable
The independent variable CEO base pay ratio is defined as the ratio of CEO base pay to the average base pay of other ASAE charitable organization CEOs.This compensation measure does not include bonuses or other forms of compensation because 54% of the CEOs had no additional compensation and of those that did report additional compensation, the additional amount was minimal (median $7,600; mean $15,710) in comparison with the average base salary, which ranged from $28,500 to $271,382 with a median and mean over $160,000.This measure is similar to the within-sample measure used by Ezzamel and Watson (1998).Rather than empirically estimating a pay anomaly, however, we used a comparison between base compensation and average sample compensation.We operationalized this variable differently than Ezzamel and Watson (1998) because financial performance and board characteristics are unavailable.Further, while CEO referents of small organizations often occupy multiple roles as part of the top management team such as CEO and CFO (Siegel & Hambrick, 2005) or a member of the board of directors (O' Reilly III, Main, & Crystal, 1988), these social comparisons may be inappropriate for some organizations (Gomez-Mejia & Wiseman, 1997) such as 501(c)(3) organizations.

Control Variables
To control for organizational and CEO-level factors that may influence performance, we include control variables for organization budget, size, noncompete agreements, geographic scope, primary operating sector, CEO tenure, and CEO gender.To control for financial resources available to the organization, we included the natural logarithm of annual budget, a value that was reported by organizations to the question "What is the total annual budget for your organization (excluding for-profit subsidiary or foundation, if any)?"The natural logarithm of organization employees was used to control for firm size effects.It was measured using the question, "How many full-time staff or full-time equivalents (FTE) are employed in your organization (excluding for-profit subsidiary or foundation, if any)?" Noncompete, captured by the question, "Is there a noncompete clause in a written contract or a non-compete understanding in a more informal arrangement to prevent conflicts of interest with other organizations?" was also a binary variable where CEOs with formal or informal noncompete agreements were coded as "1" (34%) and CEOs without such agreements were coded as "0" (66%).Noncompete agreements are used to limit employee mobility with the purpose of limiting knowledge transfer to competitors (Marx, Strumsky, & Fleming, 2009).Geographic scope is again a binary variable coded as "1" when the organization operates both within the United States and in at least one other country and was coded as "0" when the organization operates only in the United States.The majority of organizations (63%) in our sample operate internationally.Geographic scope is included in our analyses because it has been shown to influence CEO compensation (Carpenter, 1998).
We also control for organizations' primary subject area.The ASAE survey our data are drawn from includes 30 possible subject areas.Many of these subject areas, however, are closely related (e.g., law/government and law enforcement), so we combined categories into the following subsets, which we include in our analyses: wholesale (business, wholesale), manufacturing (automotive, equipment, raw materials), construction, services (administration, education, financial, healthcare, hospitality, international, labor, law, law enforcement, real estate, social welfare), agriculture, technology (architecture/engineering, energy, environment, scientific, telecommunications), and other (culture, fraternal/religious, sports).We included these subject areas because researchers have demonstrated that industry may be associated with compensation levels and structure (Gerhart & Milkovich, 1990), and we take these subject areas to be rough proxies for industries.
CEO tenure was measured as the number of years the CEO had held that position in the organization and ranged from under one year to 31 years.We include this as prior research, which has demonstrated that longer tenure may allow CEOs to build their influence within the organization and use that influence to gain a more preferable compensation package (van Essen et al., 2015).Last, we included CEO gender because research has demonstrated that female CEOs may be undercompensated relative to men (Mohan & Ruggiero, 2007).CEO gender was coded as "1" if the CEO was male (58%) and "0" if the CEO was female (42%).For the noncompete variable, if the data were missing, we coded this as "0" because we assumed that these items signal some level of CEO quality and that organizations would want to report factors associated with CEO quality.If data were missing for any other variable, the case was not included in our analyses.

Results
Ordinary least-squares regression was used to analyze the data.Moderation was tested by creating interaction terms (Baron & Kenny, 1986), and we centered interaction terms following suggested best practices (Aguinis & Gottfredson, 2010) to minimize multicollinearity concerns and aid in the interpretation of the results (Aiken & West, 1991).We tested for the presence of multicollinearity in our analyses by examining variance inflation factors (VIF); however, no evidence of multicollinearity was found.Table 1 presents means, standard deviations, and correlations for our data.The average organization in our sample served 14,974 members, had 35 full-time employees, an annual budget of $1.5 million, and had a financial efficiency score of 0.61 (meaning the average organization spent 61% of their annual budget on programing, and the remaining 39% on administrative costs).All six models were statistically significant and accounted for 36% to 60% of unexplained variance.Hypotheses 1 and 2 proposed that CEOs with less compensation than their peers will have a positive relationship with organizational performance.Results for the dependent variable individuals served are presented in Table 2, while results for the dependent variable efficiency are shown in Table 3.In Model 1 (Tables 2 and 3), we enter our control variables.In Model 2 (Tables 2 and 3), we introduce our independent variable CEO base pay ratio.In Table 2, the coefficient relating CEO base pay ratio to individuals served is negative and statistically significant.This means that the number of individuals served is higher when CEO relative pay is lower, supporting Hypothesis 1. Table 3 shows that the coefficient for CEO base pay ratio is negative but not statistically significant for financial efficiency, thus rejecting Hypothesis 2.
Hypotheses 3 and 4 proposed that the organization's annual budget would moderate the relationship between CEO relative pay and organizational performance such that the relationship would be stronger for organizations with a smaller annual budget.In Model 3 (Table 2), the relationship between this interaction (CEO base pay ratio and annual budget) and individuals served is positive and statistically significant.This interaction is more clearly seen in Figure 1, which shows that, in an organization with a smaller annual budget, having a CEO with lower relative pay is related to higher organizational performance, as measured by the number of individuals served.Thus, Hypothesis 3 is supported.Results for the relationship between this interaction (CEO base pay ratio and annual budget) and financial efficiency are shown in Model 3 (Table 3).This regression coefficient is positive and statistically significant.Figure 2 shows this interaction.CEOs with higher relative pay have lower financial efficiency when their organization has a smaller annual budget, supporting Hypothesis 4.  In Hypotheses 5 and 6, firm size (as measured by the number of full-time employees) is proposed to moderate the relationship between CEO relative pay and organizational performance such that the relationship will be stronger for smaller organizations.The interaction term in Model 3 (Table 2) is negative and statistically significant.Figure 3 shows that this interaction between CEO pay ratio and organizational size is rather complex.When the organization is larger (more employees), CEOs with a higher pay ratio lead firms that serve fewer individuals.However, when the organization is smaller, CEOs with a higher pay ratio have higher performance, as measured by the numbers of individuals served.Thus, Hypothesis 5 is not supported.In Model 3 (Table 3), the regression coefficient for this interaction (CEO pay ratio and firm size) predicting financial efficiency is negative and statistically significant.The interaction plot in Figure 4 shows that financial efficiency is lower when CEO pay is higher in larger firms.However, financial efficiency is higher when CEO pay is higher in smaller firms.Thus, Hypothesis 6 is not supported.
In addition to our hypothesized relationships, a few interesting relationships were observed with the control variables.First, in the models with the dependent variable individuals served, manufacturing and agricultural industries demonstrated significantly negative relationships in all three models, possibly suggesting that trade and professional associations in both industries tend to have smaller memberships.However, this association becomes insignificant when interacting variables of organization's size and base pay are introduced in Model 3. Geographic scope has a positive but weaker effect, implying that membership grows with geographic expansion.Even more interesting is the effect of the CEO gender on individuals served.The results suggest that female CEOs positively affect membership numbers.Regression results on efficiency do not support the same conclusions.Manufacturing has the same negative (but much weaker) effect on efficiency of organizations, and other control variables do not reach statistical significance to warrant similar conclusions.

Discussion
Using a sample of 501(c)(3) trade and membership associations, this research examined the relationship between executive compensation and organizational performance from the perspective of CEO social comparisons rather than the more typical agency theory.Although trade associations and scientific do not constitute the core of charitable nonprofits one typically thinks of when talking about the nonprofit sector, these organizations provide important public benefit directly or indirectly through educating the public about advances in their respective fields, often setting ethical and safety standards.
Our purpose was to examine how nonprofit CEOs who know they earn less than their peers may influence organizational performance.Our empirical results confirm that CEO relative pay influences two measures of organizational performance, i.e., the number of individuals served by the organization (nonfinancial performance) and organization efficiency (financial performance).While it makes intuitive sense that an organization's financial performance would improve if a CEO were paid less, our finding that CEOs who earn less than their peers might lead organizations to higher levels of nonfinancial performance was more intriguing.It seems that servant leadership does explain these results better than agency theory would, even in not "purely" charitable nonprofits.This finding is important because trade associations and professional societies are commonly considered to embody the values of the for-profit industries they serve.However, as our study showed, professional associations also embrace the more traditional values of the nonprofit sector, such as stewardship and accountability, so that the opposite of conventional wisdom can be true.Our results show that the pay-performance relationship in trade associations resembles socially oriented nonprofits, like art organizations and libraries, more so than revenue-driven hospitals and insurance organizations.On the other hand, our findings can also be interpreted using social comparison theory, where lower compensation levels can be used as a strategy to compete with others for a more favored and positive organizational image in the market.
We also demonstrated that contextual factors of firm size (measured as both annual budget and number of full-time employees) can influence this relationship.Our results were in line with past research suggesting that different subsectors of the nonprofit sector yield different patterns of behavior.However, even more nuanced differences can be observed when looking at organizational size.Trade and professional associations with larger annual budgets and smaller staff size react to a change in relative CEO pay by increasing their organizational performance and financial efficiency if CEO pay ratio is lower.As a result, these nonprofit associations seem to react in a more selfless manner than associations with smaller annual budget and larger staff.
Conversely, associations that are smaller in annual budgets but larger in staff size serve fewer stakeholders and are less efficient when relative CEO's pay is comparatively higher.Organizations with a bigger staff size routinely will have larger overhead, so the negative association with financial efficiency is not surprising.However, the fact that organizations with larger staff also serve comparatively fewer individuals is counterintuitive and warrants further investigation.One possible explanation could be that CEOs of such organizations are inefficient with their resources, which manifests itself in their smaller membership size.
Another interesting finding that deserves further attention is that CEO gender was related to an increase in the number of individuals served but not to our measure of financial efficiency.Female CEOs on average serve in smaller nonprofits, and their salaries lag behind their male counterparts even in the nonprofit sector (LeRoux & Langer, 2016), which could possibly skew the relationship with the financial performance measures.Further, the results of the relationship between CEO gender and financial performance are generally mixed, supporting our inconsistent conclusion (Pillemer, Graham, & Burke, 2014;Peni, 2014).Additional analysis is needed to explain this relationship.However, it can be suggested that female leaders focus their attention on expanding associations' reach, which requires an expansion in overhead resources, such as fundraising, employee salaries, and benefits, and these expenditures do not directly translate into financial efficiency gains.
Our findings also have practical relevance for nonprofit stakeholders, boards of directors, and CEOs.Large nonprofit CEO salary increases in recent decades have piqued public interest (Frydman & Saks, 2010), so much so that the state of New York has placed a salary cap on nonprofits that receive state funding.In light of this, understanding when higher or lower relative CEO compensation can benefit nonprofit performance is important.Specifically, nonprofit CEOs who sacrifice monetary rewards by having a base salary lower than their average peer CEO tend to lead organizations with superior performance, especially if the organization has a smaller annual budget (the mean annual budget in our sample was $1. 5 million).
Considering our findings, we do not suggest that a reduction in CEO compensation of professional associations will result in a direct and immediate increase in organizational financial efficiency and the number of individuals served.As our findings show, these connections are contextual and will also depend on a number of other factors.Further, the common concerns that large nonprofits are becoming more business-like, especially in the area of employee incentives and compensation, do not apply to the subset of professional associations in this study.One potential implication of this finding could be that the managerial compensation issues could be more effectively dealt with if focused on intrinsic, rather than exclusively extrinsic, incentives.

Limitations and Future Research
This study has several limitations, which we hope can provide avenues for future research.First, we used cross-sectional data as opposed to time-series data and utilized regression analysis, meaning that we cannot make claims of causality, and the pattern of results could in fact be in the reverse direction.Unfortunately, time-series data were not available for this study.Second, although we consider our focus on nonprofit rather than for-profit organizations, which is a strength of the paper, we used data from a compensation survey of a specific set of nonprofits, so results may not be generalizable to other types of nonprofit organizations.Relatedly, our analysis did not account for volunteers within each nonprofit, as this data were not available.While we believe that membership organizations may be different from more traditional nonprofits in that they may rely less heavily on volunteers, we assume that variance in volunteers or volunteer hours would alter our outcome variable measuring the number of individuals served.However, recent research finds that, even with for-profit organizations, the type of CEO compensation promotes more socially responsible decisions (Kang, Chiang, Huangthanapan, & Downing, 2015), and CEO identity as a philanthropist also has an impact on corporate philanthropy (Dennis & Buchholtz, 2009).Thus, future research should continue to examine multiple measures of organization performance, such as the number of individuals served or money spent providing a public service.Recent research of nonprofit hospitals finds that higher CEO compensation is related to more subjective performance measures such as patient satisfaction but not more objective performance measures of patient health like mortality rates and readmission rates (Joynt, Le, Orav, & Jha, 2014).Understanding how CEO pay has an impact on nonprofits is especially of interest given recent nonprofit financial scandals, including theft, fraud, and embezzlement (Stevens & Flaherty, 2013).Further examination of the traits, values, and context of organizational power (Williams, 2014) that can

Figure 1 .Figure 2 .
Figure 1.Interaction Between CEO Relative Pay and Annual Budget on Individuals Served

Figure 3 .Figure 4 .
Figure 3. Interaction Between CEO Relative Pay and Firm Size on Individuals Served

Table 3 .
Regression Results for Financial Efficiency