Spending Up the Ranks? The Relationship Between Striving for Prestige and Administrative Expenditures at U.S. Public Research Universities

ABSTRACT

Despite occupying a central position in contemporary U.S. higher education discourse, empirical research on administrative costs is limited. The purpose of this study was to extend existing research on administrative spending in higher education by empirically examining whether recently shifting to research university status in the Carnegie Classification influences administrative costs. Informing the analysis was a theoretical framework consisting of neo-institutional theory and the revenue theory of costs. The study examined 164 public research universities between 2004 and 2012 using a a pooled regression model with Driscoll-Kraay standard errors and a first-order autoregressive (AR1) lag. Results showed that shifting to research university status had a significant, positive influence on administrative spending at public research universities. Nevertheless, the influence of reclassification on administrative spending dissipated over time and to the point where the difference was no longer statistically significant. Importantly, results also showed that state appropriations and tuition revenue were positively associated with administrative spending, while enrollment was negatively associated with administrative spending. These results have important implications related to understanding and managing administrative spending among U.S. public research universities

KEYWORDS:

• Administrative
• administration
• spending
• cost
• expenditure
• institution
• prestige

Notes

1. We use the terms “spending,” “costs,” and “expenditures” interchangeably in this article. Additionally, we use the term “costs” in a colloquial sense rather than a technical sense (i.e., it is not inclusive of opportunity costs).

2. If the UNITID was used for this sample of institutions, then 155 rather than 164 institutions would have been in the final analytic because some were “parent” institutions to which others reported their financial data.

3. In our study, 74% of the institutions were part of systems with systemwide governing boards. These boards determine tuition policy, including tuition-setting authority. Consequently, changes in administrative expenditures are unlikely to influence changes in tuition revenue. In our statistical model specification, the results of an endogeneity test (that was fully robust to serial correlation, heteroscedasticity, and cross-sectional correlation) indicated that tuition revenue per FTE enrollment was not endogenous.

4. Because the number of time periods was substantially less than the number of institutions, time dummy variables rather than a continuous time variable were also more appropriate. Dummy variables also helped to detect nonlinearities.

5. In addition to the assumption that there is no relationship between the independent variables and the idiosyncratic error (ε), a random-effects model assumes there is no relationship between the independent variables and any unobserved group (e.g., institution) effect.

6. It should be noted that a pooled OLS regression model assumes a common intercept (α) across all institutions.

7. These results are similar to a pooled OLS model with an AR1 lag and D-K standard errors that included expenditure and revenue variables based on percentages instead of expenditure and revenue variables per FTE.

8. This figure was based on the results of the pooled OLS model with an AR1 lag and D-K errors that also included interaction terms reflecting change in the Carnegie Classification and year dummy variables.