WordPress.com



AbstractWidespread attention to college tuition and student loan debt has resulted in increasing scrutiny of high levels of compensation for college and university administrators. Prior research has sought to identify a “pay for performance” relationship in executive compensation, but discovered no clear link between presidential salaries and performance measures. This study proposes U.S. News & World Report college rankings as a highly meaningful performance metric and employs a fixed effects regression model to determine the relationship between college rankings and presidential salary. We find a significant relationship between rank and presidential salary at public universities, but not at private universities and liberal arts colleges, consistent with an academic capitalism model. Keywords: U.S. News Rankings, College Presidents, Presidential Compensation, Pay-for-Performance, Board of Trustees, Academic CapitalismIntroductionAnyone who has worked in the academe, as well as many students, have seen the headlines. “Platinum Pay in Ivory Towers,” wrote Bruni (2015) in an editorial in the New York Times. The same newspaper printed, “Salaries of Private College Presidents Continue to Rise, Chronicle survey finds,” (Saul 2015). Public college and university presidents have not been immune to criticism. “The Best-Paid Public College President in the U.S. Makes $1.55 Million a Year,” reported Mulhere (2017). “Benefits Grow for Public University Presidents, Survey Finds,” printed the New York Times (Lewin 2015). Clearly, presidential compensation at colleges and universities has struck a nerve with the public.In 2008, fewer than 10 college presidents earned more than a million dollars. In 2015, the latest year where data is available, 58 presidents (including 10 public university presidents) earned over $1 million, up from 39 the year before. As the compensation of college presidents has grown, so has tuition as well as student debt. College loan balances almost doubled in the last decade from $833 billion in 2007 to $1.4 trillion in 2017 (Dickler 2017). In addition, public universities have still not recovered from years of cuts resulting from the Great Recession. According to the Center on Budget and Policy Priorities, state funding for public two- and four-year colleges was nearly $10 billion below the amount just prior to the recession (Mitchell et al. 2016). This convergence of factors has resulted in extensive coverage of, and controversy over, the compensation of college and university presidents.Cotton (2012) argues that high levels of compensation for college presidents are warranted given the demands of their positions as well as their relative scarcity. He also argues that presidents of higher education institutions are actually underpaid when compared to the chief executive officers of large private corporations. Nevertheless, a new 21 percent tax on annual compensation in excess of $1 million paid to any of a nonprofit organization’s five highest-paid employees as part of the 2017 Republican tax plan (Harris and Bauman 2017) suggests not everyone shares this belief. Nevertheless, a pay-for-performance relationship between compensation and improved performance may justify the handsome rewards earned by presidents of higher education institutions. In this article, we examine the role of one measure of performance (though an imperfect and controversial one), U.S. News & World Report (USNWR) rankings on the compensation of college and university presidents. Since the 1990’s, U.S. News & World Report has published an annual ranking of American colleges and universities. These rankings command significant attention among students, faculty and administrators alike. The USNWR rankings influence the quality of students and faculty that a university can attract, and reinforce competitiveness in attracting research funding and other forms of investment. The substantial public relations impact produced by USNWR rankings suggest that university administrators and trustees closely monitor the rankings (Ehrenberg 2000).The impact of USNWR rankings justifies examination of their possible link with executive compensation for college and university presidents. Demonstrating a relationship between USNWR rankings and presidential salary would indicate that presidential salaries in higher education are linked to an important indicator of institutional performance, albeit one that is less concrete and more subjective and provocative than the other indicators previously studied. Establishing this connection may suggest that university trustees do value performance when determining executive compensation, especially if that performance is demonstrated in a marketable way.Our article begins with a review of the literature on the compensation of college and university presidents and USNWR rankings. Our theoretical framework follows this section. Section four presents our data. We present our regression model in section five. Section six reports results. The final section concludes and ties our results to theory.Literature ReviewPresidential CompensationA few constants hold in the research of college and university presidential pay. Research university presidents earn more than presidents at four-year institutions (Ehrenberg et al. 2001; Pfeffer and Ross 1988; Saunders 2007; Tang et al. 2000; Huang and Chen 2013; Langbert and Fox 2013). Presidents at larger institutions as measured by enrollment on average earn more than presidents at smaller institutions (Cheng 2014; Ehrenberg et al. 2001; Henry 2015; Pfeffer and Ross 1988; Saunders 2007; Monks 2007; Huang and Chen 2013; Galle and Walker 2014; Gordon and Fischer 2014), as do presidents at larger institutions as measured by revenues (Huang and Chen 2013; Monks 2007; Bai 2014; Langbert and Fox 2013). The longer a president has served, the higher his or her income is (Ehrenberg et al. 2001; Pfeffer and Ross 1988; Sorokina 2003; Galle and Walker 2014; Langbert and Fox 2013; Bartlett and Sorokina 2005). On average, leaders of private universities earn more than leaders at public universities (Monks 2007; Pfeffer and Ross 1988). Additionally, leaders of religiously affiliated institutions earn less than leaders in other types of institutions (Galle and Walker 2014; Saunders 2007; Bastedo et al. 2014; Langbert and Fox 2013)High levels of compensation for college and university presidents can perhaps be justified if there is evidence of a pay for performance relationship with compensation. For example, as the president is often called the “fundraiser-in-chief” (Hodson 2010), then a legitimate case can be made that presidents who can bring in more dollars should earn more. Nevertheless, the research is mixed on whether or not such a relationship is the case. Bartlett and Sorokina (2005), Saunders (2007), Cheng (2014), and Henry (2015) find a positive relationship between endowment size and compensation. Sorokina (2003), however, does not find a significant link between alumni giving rate and compensation. Ehrenberg et al. (2001), likewise, did not find a significant relationship between endowment or gifts and compensation. A few studies even find a negative relationship between the size of the endowment (or the share of alumni contributors) and presidential compensation (Huang and Chen 2013; Langbert and Fox 2013).The average SAT (or ACT) score of the institution is used as a measure of selectivity and prestige in the academe (Brewer et al. 2005) and hence may affect presidential compensation. Sorokina (2003), Bartlett and Sorokina (2005), Huang and Chen (2013), Langbert and Fox (2013), and Bastedo et al. (2014), all find that higher SAT scores are associated with higher compensation. However, this result may be driven by selection bias. Studies that use time-series approaches to control for unobserved variables, e.g., Ehrenberg et al. (2001), Cheng (2014), and Gordon and Fischer (2014), do not find this link, nor does Tang et al. (2000) who use a cross-sectional analysis.Another potential pay-for-performance link for college presidents is student outcomes. The empirical evidence for this link is also weak. Sorokina (2003) finds no significant relationship between freshman retention rates and presidential pay, a result supported by Langbert and Fox (2013), but not supported by Bartlett and Sorokina (2005). Likewise, Cheng (2014) does not find a significant relationship between graduation rates and presidential compensation. Gordon and Fischer (2014) even find results that suggest improvements in graduation rates result in lower presidential compensation.USNWR RankingsSince 1983, the annual rankings of American colleges and universities reported by the USNWR have fulfilled an important role in higher education. Students use the rankings to inform their decisions when applying to colleges and choosing where to enroll. In a poll of college-bound students conducted in 2012, two-thirds reported taking college rankings into account when deciding upon a college. Additionally, eighty-five percent of students with SAT’s 1300 or above relied on the rankings (O'Shaughnessy 2013). Aware of their influence on attracting and enrolling quality students and faculty, university administrators and trustees make decisions to maximize their performance in the rankings (Ehrenberg 2000). USNWR rankings are not the only rankings of postsecondary institutions, though they are the most well-known. The USNWR rankings are geared towards students and their parents and are based to a large extent on prestige (Bowman and Bastedo 2009). Global rankings like The Times Higher Education’s World University Rankings and Academic Rankings of World Universities are more interested in research output (Cantwell and Taylor 2013).The media has speculated that trustees and presidents may be overly concerned with USNWR rankings, and may manipulate various aspects of their institutional practices and reporting to “game the system” (Butler 2009; De Vise 2011). Ehrenberg (2002) reports that while university presidents publicly proclaim that USNWR rankings do not adequately measure the quality of their institutions, they endeavor to maximize their rankings from behind the scenes. He shares numerous anecdotes, including colleges that implemented public relations damage control after falling in the rankings and colleges that engaged in strategic planning to bolster their rankings. Because the formulae of USNWR’s rankings are well known, schools attempt to manipulate their policies to score higher on key measures, such as admissions selectivity, peer evaluations from competitor institutions, financial resources per student, and full-time faculty salaries. For example, Ehrenberg observes that college administrators have been incentivized to solicit admissions applications from unqualified students to increase selectivity, to invest in expensive marketing and promotional materials sent to other academics, to spend more than cut costs (further driving up tuition), and to replace tenure-track faculty with adjuncts to raise the salaries of remaining faculty. While the rankings are highly subjective, and prone to manipulation as reported in Ehrenberg (2002), they have demonstrated a significant impact on a number of admissions indicators. Monks and Ehrenberg (1999) find that negative changes in USNWR rankings adversely impacted acceptance and matriculation rates and quality of incoming freshman classes. These effects, in turn, forced colleges to offer higher levels of financial aid to attract students, which decreased operational revenue. Bowman and Bastedo (2009) similarly find that appearing on the “front page” of the U.S. News rankings by moving into the Top 50, provided a substantial improvement in admissions indicators. These institutions saw increases in the share of incoming freshmen who graduated in the top 10 percent of their high school class as well as declines in acceptance rates and increases in applications, which are consistent with Meredith (2004). Bowman and Bastedo (2009) determine that the effect was stronger for highly ranked national research universities. Weisbrod et al.’s (2008) more limited analysis of the top 25 USNWR institutions however, did not find changes in enrollment or applications.Weisbrod et al. (2008) also find that schools that declined in ranking lowered their tuitions the following year. In addition, Zhe Jin and Whalley (2007) find that after changes in 1990 which expanded the scope of USNWR rankings, inclusion in the new rankings delivered a 6.8 percent increase in state funding per student. Finally, Lee et al. (2014), conclude there were significant relationships between USNWR rankings and graduation rate, per student expenditures, alumni giving, acceptance rate and class sizes under 20. Collectively these findings indicate that USNWR rankings have substantial effects on colleges and universities, especially when it comes to financial resources and student quality.The Effect of Rankings on Presidential CompensationThe high visibility of these rankings as well as the theory presented below suggest a link between presidential compensation and USNWR rankings. Several studies have indeed looked at this relationship. Sorokina (2003) finds that USNWR ranking (in the form of changes in tiers) was a significant predictor of presidential salary at liberal arts colleges. Tang et al. (2000) also find a strong relationship between compensation and institutional prestige, linking presidential salaries with USNWR rankings, and suggest these rankings may be an important measure of institutional performance at private institutions. Weisbrod et al. (2008) likewise estimate that an increase of one position in USNWR ranking in 2005 was associated with an approximately $2,037 increase in presidential compensation the following year for presidents at Division I-A schools. All of these studies were cross-sectional analyses. Bai’s (2014) time-series analysis, however, finds no relationship between ranking and presidential compensation at research universities ranked between 1 and 100. In addition, Saunders (2007) finds little evidence that USNWR tiers affected compensation of school presidents in the 2004-2005 academic year. These conflicting results justify a more detailed investigation into a link between USNWR rankings and college presidential compensation.Our study is different from these studies in several ways. First, we use a more recent dataset on higher education (the most recent year is 2008-2009). In addition, to our knowledge, this is the first study that looks at both private and public institutions as well as liberal arts colleges. Finally, the panel nature of our data allow us to remove the influence of unobserved institutional and individual characteristics and isolate the effect of a change in USNWR ranking on the compensation of college and university presidents.Theoretical FrameworkWhy would we expect compensation to be tied to USNWR rankings? Would it differ by the type of college? In this section, we review three major theories involving college executive compensation, and their links with USNWR rankings. We should note that these theories are by no means exhaustive, nor are they mutually exclusive.Principal-Agent TheoryExecutive compensation can be understood through the labor economics perspective of the principal-agent theory. Within this framework, a “principal” or industrial firm, delivers compensation to an “agent,” or employee, in order to secure performance that promotes the objectives of the firm. This theoretical framework is the basis for the pay-for-performance model of executive compensation. Under this model, the principal rewards service that improves the performance of the firm by increasing compensation (Murphy 1999).Multiple studies investigate a pay-for-performance link in executive compensation in higher education under the principal-agent framework. Examining data from a wide spectrum of colleges and universities, Ehrenberg et al. (2001) evaluate the effect of a number of performance indicators, including individual and institutional characteristics and find weak evidence for a pay-for-performance relationship. Cheng (2014) finds no evidence for a statistically significant pay-for-performance principal-agent relationship at 128 public research universities. Bartlett and Sorokina (2005) also claim weak evidence for a pay-for-performance link, demonstrating that institutional prestige and personal characteristics had a larger effect than performance measures on executive compensation at private liberal arts colleges.Results from our study that showed presidents of all types of institutions had higher pay from improved rankings would provide support for a pay-for-performance model. This finding would suggest that the president was rewarded by the board of trustees for improving performance in the form of the USNWR rankings.Bureaucratic ModelIt remains unclear whether the “pay-for-performance” model accurately describes compensation structures in higher education. Cheng (2014) argues inconsistencies in applying principal-agent theory to higher education administration, citing Lane (2012) and Langbert (2006) who identify multiple principals, including governing boards, state government, constituent colleges, faculty, students and alumni that serve as roadblocks to maximizing performance. In addition, a long line of research such as Wittmer (1991), suggests that employees in public service sectors are less interested in pay and more interested in doing work that benefits others than private sector employees. If such is the case, we would expect presidential compensation to not be responsive to changes in measures of performance, like fundraising, or USNWR rankings.Cheng (2014) argues a more bureaucratic payment scheme that is predominantly fixed in nature, mirroring compensation models in other public and non-profit sector organizations is the best approximation of how presidents in higher education institutions are compensated. Instead, presidents receive raises like bureaucrats in regular, expected increments with no ties to performance.Academic CapitalismThis theory, articulated by Weisbrod et al. (2008) among others, holds that colleges and universities, and even public and non-profit ones, are influenced by the pursuit of revenues and are in competition with each other for those revenues. These organizations must balance the money-losing activities related to their mission (e.g. pure research, instruction of the less affluent), with money-making activities (e.g. applied research, education of the wealthy). This drive for revenues creates a market where colleges and universities compete against each other. According to Weisbrod et al. (2008), “[Colleges and universities] compete for students—sometimes as part of their educational mission but sometimes simply as revenue sources—for individuals’ donations, for governmental research grants and corporate research support, and for star athletes and ever star academics.” The data bear this competition out. Colleges and universities, even public ones, compete in a national market. Forty-three of the nation’s 50 state flagship schools have out-of-state student populations that outnumber their in-state populations. At the University of Alabama for example, in 2004, 72 percent of freshmen hailed from Alabama. By 2014, the share was 36 percent (Anderson and Douglas-Gabriel 2016).This growth in out-of-state students at public colleges and universities is a reaction to long-term reductions in state and local funding. Out-of-state students pay more tuition than their in-state peers, on average, twice as much (Ma et al. 2016). Tuition, not state aid, now represents a majority of revenues at most public institutions (Desrochers and Hurlburt 2016; Slaughter and Leslie 1997). This phenomenon has occurred despite the stated mission of such institutions to provide students in their own states a quality education at a reasonable price (McKenna 2015). As discussed previously, lower USNWR rankings have been associated with declines in the quality of student body as well as decreases in tuition (Monks and Ehrenberg 1999; Weisbrod et al. 2008). These results may be because the rankings serve as measures of quality for applicants and their families. In a competitive environment, they are a way for schools to stand out. The rankings would be especially important for public universities, which have traditionally been more well-known regionally (Weisbrod et al. 2008), to reach higher-tuition out-of-state students. Given the importance of rankings to increasing prestige and the importance of out-of-state students to public university budgets, the compensation of presidents at public institutions may be linked to changes in their schools’ USNWR rankings.DataThe structure of this paper is influenced by Ehrenberg et al. (2001). Their paper determined the effect of individual, institutional and performance characteristics on the salaries of college and university presidents between the years 1992-1993 and 1997-1998. This paper replicates their analysis with an annual data panel of the same variables over the period 2001-2002 to 2008-2009 while focusing primarily on the effect of UNSWR rankings on the compensation of college presidents, especially those of public institutions.Our dependent variable is the log of presidential compensation. Our independent variables of interest are USNWR ranking and an interaction between USNWR ranking and private university, which allows the effect of ranking to vary based on university type. To reduce omitted variable bias, we control for a series of both presidential and institutional covariates which might be correlated with both presidential compensation and USNWR rankings.The first set of presidential covariates include many personal characteristics of university presidents. Because experience likely plays a role in compensation and peer assessment of the university (and thus USNWR rankings), we control for tenure at the current presidency, whether the individual has prior experience as a president at another institution, and the length of his or her term at the prior institution. In addition, presidents in their final year of presidency may receive a bonus (a golden parachute), so we control for the final year of each president’s tenure. Finally, we control for age and gender, which are known to be associated with disparities in compensation in the labor market (Ehrenberg et al. 2001).We also include variables indicating institutional characteristics, because characteristics that differ between universities may lead to differing levels of compensation. As larger institutions may pay higher salaries, we control for total full-time enrollment. Since different institutional types may have different compensation schemes, we control for indicators of private universities and liberal arts colleges. Financial variables which affect revenues and expenditures are also likely to impact presidential compensation, so we control for the logs of endowment per student, institutional gifts per student, average faculty salary and research and development expenditures. Finally, presidents may be rewarded for increasing student quality, so we control for average freshman SAT scores (Ehrenberg et al. 2001). Average SAT was calculated as a sum of average verbal SAT score and average math SAT score. These variables may all be indicators of a pay-for-performance relationship with compensation. This paper employs data from five sources. Rankings were collected from the USNWR Ultimate College guidebook. Individual characteristics of the college and university presidents were collected from Who’s Who and Who’s Who in Academia. Presidential salary data were collected from the Chronicle of Higher Education, which publishes presidential salary data as reported by private colleges and universities on their Internal Revenue Service form 990. There is some variability in these numbers, with colleges adhering to different reporting standards in terms of perquisites and outside compensation. Nevertheless, they are considered the best available estimates of presidential compensation (Ehrenberg et al. 2001). The Chronicle also surveys public institutions to collect data on the compensation of public university presidents. These data are not directly comparable to the 990 data. The figures include slightly different values, like the value of various nontaxable benefits (Hatch 2017).To replicate the institutional variables selected by Ehrenberg et al. (2001), we collected data from two additional sources: the Integrated Postsecondary Education Data System (IPEDS)00 for institutional variables and the National Association of College and University Business Officers-Commonfund Study of Endowment for endowment data. These variables are collected annually for the same institution.There is a timing issue with using the different sources of data. Take the year 2005 for example. This 2005 year is part of two school years, 2004-2005 and 2005-2006. The 2005 USNWR ranking was published in the fall of 2004 for academic year (AY) 2004-2005. Financial data are reported to IPEDS at the end of the fiscal year at the end of June. The IPEDS data, e.g., SAT data and enrollment, are reported for the fall of the current school year (the fall of 2004). As budgeting decisions are set in the spring of the previous fiscal year (namely the spring of 2004, at the end of AY2003-2004, it is important that our analyses use the data which trustees have access to when they make compensation decisions. A president’s total compensation earned at the end of June 2005 could not be based on 2006 USNWR rank as that source of data was not yet published. Therefore, we use the 2005 rank published in the fall of 2004, the fall 2004 admissions data, and the spring 2004 financial data including salary for our analyses.Descriptive StatisticsTable 1 presents descriptive statistics for our entire sample of four-year colleges and universities. The sample we use for our regressions consists of over 900 observations, including data from 166 nationally ranked doctoral universities and liberal arts colleges, for the period between 2001-2002 and 2008-2009. Data on many of the public universities in this sample are only available for the 2004-2005 to 2008-2009 academic years. The average institution was ranked 51.6 with the average liberal arts college ranked 42.2 and the average doctoral university ranked 57.3. Presidents on average earned over $540,000, though there is considerable variation in this amount. Benjamin Ladner of American University was the top earner, receiving over $4 million in compensation in 2006. Consistent with the corporate world where women are minorities in executive positions (Soares et al. 2010), women were also a minority in our dataset, accounting for less than 20 percent of college and university presidents. In addition, most presidents were in their first position as the leader of an institution. College and university presidents during the time period of this study tended to be an older lot, with an average age of just over 60 years old. The youngest presidents were 45 years old, namely, Anthony W. Marx of Amherst College, and Holden Thorp of the University of North Carolina at Chapel Hill. Aram V. Chobanian of Boston University and Derek C. Bok of Harvard University, were the oldest presidents at 77 years of age. A little less than three-quarters of our sample is composed of private institutions. The average institution in our data had an enrollment of 13,754 students, though this number ranged from 680 students for Sweet Briar College, an all women liberal arts college, all the way up to over 160,000 students for University of Texas at Austin, an R1 research university. The average institution in our data had an average combined SAT score of 1,272. The institution with the lowest average SAT score of 1,040 was the University of California at Riverside and the institution with the highest average SAT score of 1,525 was the California Institute of Technology. Among the variables with the widest distributions are the financial variables. For example, endowments per student ranged from just over $2,600 for the University of California at Santa Cruz to over $2.25 million dollars for Princeton University, with an average endowment per student of $195,929. The average college or university in our data had $9,690 in gifts for each student, with a standard deviation of $12,522. There was also a wide dispersion in the spending on research and development. Claremont McKenna College, a liberal arts college in California, spent just over $6,000 in research and development, while the University of Texas at Austin, spent over $1.35 billion, with the average institution spending $145 million. The average institution paid faculty on average over $80,000 with a standard deviation of over $16,000. Despite these wide distributions, the highest value of our Cook’s distance calculations was 0.11, suggesting no influential outliers. We therefore retained all observations in our dataset. Estimation ModelWe are influenced by Ehrenberg et al. (2001) in the development of our model. The primary estimation equation we use in this study is presented in equation (1):lnCOMPENSATIONist=βRANKst+ζPRIVATEs+ηPRIVATEs*RANKst+γINDIVIDUALit+δINSTITUTIONALst+τt+νs+μi+εist .(1)In equation (1), the log of presidential compensation, COMPENSATION, for president i of school s in year t is a function of the USNWR rank, RANK, a dummy variable for private institutions, PRIVATE, an interaction term between these two variables, RANK * PRIVATE, a vector of individual characteristics for each president, INDIVIDUAL, like age and sex, a vector of institutional level characteristics, INSTITUTIONAL, like average faculty salary and enrollment, and an idiosyncratic error term, ε. The interaction term allows the effect of USNWR ranking on presidential compensation to vary between private and public institutions. We use fixed effects regression in this study, which we are able to utilize due to the panel nature of our data. Also in (1) are a set of fixed effects for year, τ, institution, ν, and president, μ. As discussed in Yeung and Nguyen-Hoang (2014), fixed effects models are a method for addressing selection bias. Year fixed effects capture annual common factors that affect presidential salary in all higher education institutions nationwide, e.g., inflation or overall economic condition. The Great Recession starting in late 2007, for example, an event that affected almost all higher education institutions, would be controlled for through this year fixed effect. Institutional or school fixed effects included in our regressions control for all of the observed and unobserved characteristics unique to a college or university (such as prestige) that do not change over time. In another specification, we also control for individual fixed effects in addition to school fixed effects, to hold constant all time-invariant individual factors, such as fame or charisma, that are likely correlated with USNWR ranking and presidential pay.Inclusion of individual fixed effects leads to the dropping of time-invariant variables (e.g., sex) out of the analyses. We also observe that years prior president and length of term also drop out of the regression analysis as there was very little individual-specific variation in these variables. However, a president who was observed at least twice in the dataset, and who moved to a different institution during the sample period, would be a prior president at the second institution having served in the first one. This dummy variable of the “prior president” status would, hence, not drop out of the analysis. We ran residual plots on our independent variables to test for linearity. The results suggest the final forms of our variables have linear relationships with the log of salary. We calculate Huber-White robust standard errors for all our regressions and adjust standard errors for clustering (correlations) within institutions.ResultsCross-TabulationsWe begin our analyses by calculating simple cross-tabulations of the data between our binary variables and presidential compensation to establish some basic relationships between our variables and to further explore our dataset. These analyses, reported in Table 2, of course, did not control for omitted variables as our fixed effects regressions do.While there were many more men who served as presidents in the data than women in the data, we do not, at least per these analyses, find a significant difference in compensation between male and female presidents. While men made approximately $20,000 more, the result is not significant. This appears different from the private sector, where there is a significant and persistent gender gap in executive compensation (Soares et al. 2010).Experience did appear to matter with prior presidents earning $64,820 more than first-time presidents do, a result significant at the 0.05 level. The type of institution also appears to have an impact on compensation. Private school presidents earned approximately $50,000 more than their public-school counterparts did. The biggest difference however, may be between research universities and liberal arts colleges. Presidents at research universities earned an impressive $225,359 more than presidents at liberal arts colleges.CorrelationsTable 3 presents correlations between our continuous variables with USNWR rank, and presidential compensation. Once again, these analyses are exploratory in nature and do not control for other variables as our regressions did.USNWR ranking has a negative and significant correlation with presidential compensation. It is important to remember that the top ranked school has the lowest numerical value. Therefore, this correlation suggests that higher ranked schools have presidents who earned more money. All of the other continuous variables, namely age, years as a prior president, length of term, enrollment, freshman test scores, endowment per student, gifts per student, research and development expenditures and average faculty salary, are positively and significantly associated with presidential compensation. The last variable has a particularly strong association with presidential compensation. It appears colleges and universities that paid their faculty more, also paid their presidents more.Schools with younger presidents and smaller schools tended to be lower ranked institutions. In addition, several of the other continuous variables have strong correlations with USNWR rank. Some of these variables like freshman test scores were factors contributing to the rank. Financial variables, including endowment and gifts per student and research and development expenditures, also appear to be strongly correlated with USNWR ranking. Institutions that paid their faculty more also had superior USNWR ranks.Fixed Effects RegressionsTable 4 reports the results of our fixed effects regressions, with the log of presidential compensation as dependent variable. Column I controls for institutional (school) fixed effects and Column II adds individual fixed effects. Because these two specifications control for different levels of fixed effects, the results for these two different specifications may be different.Column I uses variation in compensation from different presidents at the same institution, while Column II uses variation within the same president over time at the same institution. As seen in Table 4, the age and prior president variables are significant in Column II but not in Column I. Likewise, term length and the log of average faculty salary are significant in Column I but not in Column II. Again, these opposing results are because the variation used to estimate the independent variable coefficients are different. The within R-squared (a measure of goodness of fit for fixed effects models) for the individual + institutional fixed effects model (0.471) is larger than for the institutional fixed effects alone model (0.397). This along with the capability of controlling for additional unobserved time-invariant individual effects exhorts us to place more trust in the results in Column II. Robust results across both specifications have even greater veracity.Table 4 shows that in both columns, the coefficient on USNWR ranking is negative and significant and the interaction term between USNWR ranking and private institution is positive and significant. Because of the interaction term, the ranking variable must be interpreted conditional on whether the institution was a public or private institution. When the institution is a public university (i.e., the interaction term is zero), the coefficient on the ranking variable is the effect of ranking on compensation for public institutions. Because the top institutions have the lowest numerical ranks, a negative coefficient on the ranking variable means that presidents who saw their ranking drop made less money. Or to put it another way, improving the ranking would result in higher levels of compensation. The results in these two columns suggest that an improvement in USNWR ranking by one spot increases a public university’s presidential compensation by 1.1 to 1.3 percent. The estimate of the USNWR ranking variable is significant at the 0.01 level in both specifications, although one should note that a large sample size—as in our study—is a factor influencing statistical significance.The average public university president in our sample earned $479,116. Based on Column I, this president would earn $6,229 more a year by improving the USNWR rank by one position. If he or she can improve the ranking by 10 spots, he or she would earn $62,285 more. Based on Column II, said president would be compensated $5,270 for improving the school’s USNWR by one position and $52,703 for improving the school’s USNWR rank by 10 spots.For private institutions, the USNWR ranking variable and the interaction term with private institution must be interpreted together to calculate the effect of ranking on presidential compensation for this group of colleges and universities. Table 4 reports that the interaction term is significant at the 0.05 level in Column I and Column II. However, combined, the main term and the interaction term, in essence, cancel each other out, suggesting a very small, almost zero effect from USNWR ranking on the compensation of private college and university presidents. In Column I, we find that term length influences presidential compensation. We find that each additional year in office is associated with a 1.5 percent increase in compensation, a result significant at the 0.01 level. We see these effects as somewhat interchangeable with the effects of age. While not significant in column I, in column II we see that presidents gain a 6.6 percent increase in compensation per year of age which is significant at the 0.01 level. We uncover little evidence that institutional level variables, like enrollment, selectivity as measured by freshman tests scores, the size of the endowment, or the amount of research and development expenditures have an impact on compensation. This is likely the result of multi-collinearity with the ranking variables. Table 3 reports the USNWR ranking is highly correlated with several of the institutional variables. The ranking variable acts a composite capturing the effects of other institutional variables.The institutional-level variable with the strongest effect on presidential compensation is the average faculty salary. We find that a 10 percent increase in average faculty salary is associated with an approximately 46.1 to 54.0 percent increase in presidential compensation. The result is significant at the 0.05 level in Column I, and just misses significance in Column II, which may be the results of insufficient intra-president variation in average faculty salary or reduced power resulting from a smaller number of degrees of freedom. This finding makes sense intuitively, as faculty and administration share in the strength and weakness of the institution. A flourishing college or university may mean that raises trickle down to everyone while a college or university experiencing financial challenges may require everyone to suffer a cut.ConclusionThe findings in this study provide some evidence to support a pay-for-performance link in presidential compensation. On one hand, the model demonstrates that the U.S. News & World Report ranking, a highly visible measure of institutional performance, is linked to presidential compensation at public universities. However, this relationship does not exist at private liberal arts colleges and universities. While very few measures of institutional performance are associated with increased presidential salaries, it should be reiterated that the ranking variable is correlated with many of the measures of institutional performance. By improving selectivity in the form of SAT scores, for instance, a president is also likely to improve the institution’s USNWR ranking.These results are both consistent with and in contrast to much of the research on USNWR rankings and presidential compensation. Most of the research on this issue have been cross-sectional analyses of private institutions, and have generally found positive relationships between rankings and presidential compensation. Our study uses a time-series approach and includes public, private, and liberal arts colleges and universities and finds a positive relationship with presidential compensation, but only for public institutions.But why would ranking be a significant predictor of presidential compensation at public research universities, but not at private liberal arts colleges and universities. This could be because private universities see their reputations as well established, while public universities see the potential to gain prestige by rising in the rankings. In the words of Weisbrod et al. (2008), “Princeton does not need to ‘sell’ its college experience to the vast majority of applicants its reputation for superior undergraduate education is long established.” In contrast, the University of Nebraska-Lincoln, may be more reliant on rankings to attract students from outside its region. This is consistent with an academic capitalism hypothesis. Public institutions have seen their levels of state aid decline over time and have replaced these revenues with tuition, particularly revenue from out-of-state students, who pay significantly higher rates of tuition than their in-state peers. Public college and university presidents may be compensated for their ability to attract these out-of-state residents (through improved USNWR rankings) as well as their tuition dollars. Suggestions for Further ResearchAs discussed previously, the USNWR rankings are not the only set of academic rankings. The Times Higher Education’s World University Rankings and Academic Rankings of World Universities are based primarily on research output. At many (if not most) universities, faculty are rewarded primarily based on their research record and not on their teaching ability (Serow 2000). Also, international students are likely to pay the same rate as out-of-state students at public higher education institutions. It therefore warrants future research to examine whether U.S. students pay attention to these research-based international rankings, and whether presidents of American colleges are rewarded for improving these rankings.Conflict of Interest: The authors declare that they have no conflict of interest.Acknowledgements: The authors would like to thank Dana St. Aubin for help with data collection, the support of Distinguished Professors Award for Graduate Student Research at SUNY-Brockport, and two anonymous referees for their comments.ReferencesAnderson, N., & Douglas-Gabriel, D. (2016, January 30). Nation’s prominent public universities are shifting to out-of-state students. Washington Post.Bai, G. (2014). University president compensation: Evidence from the United States. Higher Education Studies, 4(6).Bartlett, R. L., & Sorokina, O. V. (2005). Determinants of presidential pay at national liberal arts institutions. Review of Higher Education, 29(1), 53-68.Bastedo, M. N., Samuels, E., & Kleinman, M. (2014). Do charismatic presidents influence college applications and alumni donations? Organizational identity and performance in US higher education. [journal article]. Higher Education, 68(3), 397-415.Bowman, N. A., & Bastedo, M. N. (2009). Getting on the front page: Organizational reputation, status signals, and the impact of U.S. News and World Report on student decisions. Research in Higher Education, 50(5), 415-436.Brewer, D. J., Gates, S. M., & Goldman, C. A. (2005). In pursuit of prestige: Strategy and competition in U.S. higher education. New Brunswick, N.J.: Transaction Publishers.Bruni, F. (2015, May 20). Platnium pay in ivory towers. [Editorial]. New York Times.Butler, K. (2009, June 3, 2009). How to game the college rankings: Tips from Clemson. Mother Jones.Cantwell, B., & Taylor, B. J. (2013). Global status, intra-institutional stratification and organizational segmentation: A time-dynamic tobit analysis of ARWU position among U.S. universities. Minerva, 51(2), 195-223.Cheng, S. (2014). Executive compensation in public higher education: Does performance matter? Research in Higher Education, 55(6), 581-600.Cotton, R. D. (2012). Myth: College and university presidents are overpaid. The Presidency, De Vise, D. (2011, September 3). U.S. News college rankings are denounced but not ignored. Washington Post.Desrochers, D., & Hurlburt, S. (2016). Trends in college spending: 2003-2013 - Where does the money come from? Where does it go? What does it buy? Washington, DC: American Institutes for Research.Dickler, J. (2017). Student loan balances jump nearly 150 percent in a decade. College Game Plan, Ehrenberg, R. G. (2000). Tuition rising: Why college costs so much. Cambridge, MA: Harvard University Press.Ehrenberg, R. G. (2002). Reaching for the brass ring: The U.S. News & World Report rankings and competition. Review of Higher Education, 26(2), 145-162.Ehrenberg, R. G., Cheslock, J. J., & Epifantseva, J. (2001). Paying our presidents: What do trustees value? Review of Higher Education, 25(1), 15-37.Galle, B., & Walker, D. I. (2014). Nonprofit executive pay as an agency problem: Evidence from U.S. colleges and universities. Boston University Law Review, 94, 1881-1934.Gordon, G., & Fischer, M. (2014). Public university presidential compensation: Performance, cost efficiency and spending choices. Journal of Public Budgeting, Accounting & Financial Management, 26(4), 557-583.Harris, A., & Bauman, D. (2017, December 19). New tax law takes aim at higher education’s millionaires club. Chronicle of Higher Education.Hatch, J. (2017). Re: Question about compensation data. In R. Yeung (Ed.).Henry, T. F. (2015). Are public university executives paid for their performance? In C. R. Lehman (Ed.), Sustainability and Governance (pp. 161-181, Advances in Public Interest Accounting, Vol. 18): Emerald Group Publishing Limited.Hodson, J. B. (2010). Leading the way: The role of presidents and academic deans in fundraising. New Directions for Higher Education, 2010(149), 39-49.Huang, Y. S., & Chen, C. R. (2013). Are college chief executives paid like corporate CEOs or bureaucrats? Applied Economics, 45(21), 3035-3043.Lane, J. E. (Ed.). (2012). Agency problems in higher education administration (The Organization of Higher Education: Managing Colleges for a New Era). Baltimore, MD: Johns Hopkins Press.Langbert, M. (2006). How universities pay their presidents. Academic Questions, 19(2), 67-81.Langbert, M., & Fox, M. (2013). The compensation and benefits of private university presidents Journal of Academic Administration in Higher Education, 9(2), 45-56.Lee, G., Sanford, T., & Lee, J. (2014). Variables that explain changes in institutional rank in U.S. News & World Report rankings. KEDI Journal of Educational Policy, 11(1), 27-47.Lewin, T. (2015, June 7). Benefits grow for public university presidents, survey finds. New York Times.Ma, J., Baum, S., Pender, M., & Welch, M. (2016). Trends in college pricing 2016. Trends in Higher Education Series. New York: College Board.McKenna, L. (2015, October 15). The allure of the out-of-state student. The Atlantic.Meredith, M. (2004). Why do universities compete in the ratings game? An empirical analysis of the effects of the U.S. News and World Report college rankings. Research in Higher Education, 45(5), 443-461.Mitchell, M., Leachman, M., & Masterson, K. (2016). Funding down, tuition up. Washington, D.C.: Center on Budget and Policy Priorities.Monks, J. (2007). Public versus private university presidents pay levels and structure. Economics of Education Review, 26(3), 338-348.Monks, J., & Ehrenberg, R. G. (1999). The impact of U.S. News & World Report college rankings on admission outcomes and pricing policies at selective private institutions. NBER Working Paper Series No. 7227.Mulhere, K. (2017). The best-paid public college president in the U.S. makes $1.55 million a year. CNN Money, Murphy, K. J. (Ed.). (1999). Executive compensation (Handbook of Labour Economics). Amsterdam: North-Holland.O'Shaughnessy, L. (2013). College rankings more important than ever. CBS Moneywatch, Pfeffer, J., & Ross, J. (1988). The compensation of college and university presidents. [journal article]. Research in Higher Education, 29(1), 79-91.Saul, S. (2015, December 6). Salaries of private college presidents continue to rise, Chronicle survey finds. New York Times.Saunders, K. T. (2007). Salary study of college presidents and faculty: Are salaries for institutions in the council of christian colleges and universities different from other private institutions? Christian Business Academy Review, Serow, R. C. (2000). Research and teaching at a research university. Higher Education, 40(4), 449-463.Slaughter, S., & Leslie, L. L. (1997). Academic capitalism: Politics, policies, and the entrepreneurial university. Baltimore: Johns Hopkins University Press.Soares, R., Combopiano, J., Regis, A., Shur, Y., & Wong, R. (2010). 2010 Catalyst Census: Fortune 500 Women Board Directors. New York: Catalyst.Sorokina, O. V. (2003). Executive compensation: The case of liberal arts college presidents. Issues in Political Economy, 12.Tang, T. L.-P., Tang, D. S.-H., & Tang, C. S.-Y. (2000). Factors related to university presidents' pay: An examination of private colleges and universities. Higher Education, 39(4), 393-415.Weisbrod, B. A., Ballou, J. P., & Asch, E. D. (2008). Mission and money: Understanding the university. New York: Cambridge University Press.Wittmer, D. (1991). Serving the people or serving for pay: Reward preferences among government, hybrid sector, and business managers. Public Productivity & Management Review, 14(4), 369-383.Yeung, R., & Nguyen-Hoang, P. (2014). Fixed-effects models. In D. Brewer, & L. O. Picus (Eds.), Encyclopedia of education economics and finance (pp. 350-352). Los Angeles: Sage.Zhe Jin, G., & Whalley, A. (2007). The power of information: Do rankings affect the public finance of higher education? National Bureau of Economic Research, Working Paper No. 12941.Table 1 Descriptive statistics?MeanStandard DeviationMinimumMaximumNDependent VariablePresidential compensation541,806.80323,680.3013,683.004,270,665.00930Independent VariablesU.S. News Ranking45.3633.401.00124.00889Individual-level VariablesFemale0.190.390.001.00896Age60.255.4345.0077.00894Prior president0.230.420.001.00897Years prior president1.593.530.0026.00896Length of term6.534.730.0023.00896Final year of presidency0.070.260.001.00897Institutional-level VariablesPrivate institution0.710.460.001.00941Liberal arts college0.380.480.001.00941Full-time equivalent enrollment13,754.0617,023.12680.46163,777.50941Average freshman test scores1,272.19107.421,040.001,525.00941Gifts per student9,690.2812,522.540.00185,111.90941Research and development expenditures 145,000,000 210,000,000 6,125 1,360,000,000 937Average faculty salary82,928.3416,164.5951,596.90166,697.60941Table 2 Cross-tabulations of average presidential compensation?NMean?SexMale696$563,317Female178$541,548Difference in means$21,769T-test0.795Prior presidentYes194$609,418No681$544,598Difference in means$64,820T-test-2.453*Type of institutionPrivate876$527,172Public273$477,441Difference in means$49,730T-test-2.356*Emphasis of curriculumResearch university684$606,559Liberal arts college465$381,200Difference in means$225,359T-test?13.178**Note: *significant at 0.05; **significant at 0.01.Table 3 CorrelationsU.S. News rankPresidential CompensationU.S. News rank1.000-0.101*Presidential compensation-0.101*1.000Age0.212*0.110*Years prior president0.0540.099*Length of term0.0550.116*Full-time equivalent enrollment0.199*0.165*Freshman test scores-0.864*0.232*Endowment per student-0.569*0.135*Gifts per student-0.539*0.129*Research and development expenditures-0.231*0.257*Average faculty salary-0.502*0.452*Note: *Significant at 0.05 level.Table 4 Fixed Effects Regression Results, Log of Presidential Compensation as Dependent Variable?I?II?U.S. News Ranking-0.013**-0.011**(2.98)(2.84)U.S. News Ranking * Private Institution0.014*0.011*(2.31)(2.16)Individual-level VariablesFemale0.104(1.47)Age-0.0000.066**(0.05)(3.44)Prior president0.1920.063*(1.55)(1.98)Years prior president-0.005(0.29)Length of term0.015** 0.024(2.72) (1.44)Final year of presidency0.0400.061(0.84)(1.11)Institutional-level VariablesLog of full-time equivalent enrollment-0.062-0.640(0.20)(1.78)Average freshman test scores0.000-0.001(0.00)(1.09)Log of endowment per student0.019-0.045(0.25)(0.56)Log of gifts per student-0.008-0.020(0.28)(0.57)Log of research and development expenditures0.0210.011(0.98)(0.57)Log of average faculty salary0.540*0.461(2.19)(1.74) Constant6.729*10.680**(1.99)(3.62)N831831Within R-Squared0.3970.471Fixed Effects?InstitutionalInstitutional +IndividualNotes: All regressions include year fixed effects. Absolute value of t-statistics calculated using Huber-White robust standard errors adjusted for clustering by institution presented in parentheses. *significant at 0.05; **significant at 0.01. ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download