Do Elite Universities Overpay Their Faculty?

This paper measures the relation between faculty salaries (net of faculty quality) and university or college prestige. The authors find no evidence that more prestigious institutions pay premiums above the competitive salary for the quality of the faculty they hire. Indeed, using an AKM (Abowd et al., 1999) model, The authors find little evidence of any institution effect on salaries, although salaries are higher in urban areas. The absence of institution effects in the AKM model is striking. It implies that, aside from random factors, faculty would receive the same salary at any university. The authors authors find it implausible that Oakland University would be willing to match the salaries Stanford pays its tenure-stream faculty. 

The authors draw on the Survey of Doctorate Recipients (SDR), a panel survey of individuals with U.S. doctorates in fields covered by the National Science Foundation. Thus,the results apply to STEM and the social sciences but not necessarily to the humanities or faculty with professional degrees. The authors merge the SDR with IPEDS institutional data and rankings of colleges and universities. The authors begin by applying a standard AKM model. The variance of the institution fixed effects is as little as .007, depending on the correction, and even lower if te authors apply the Kline et al. (2020) leave-one-out correction to the narrower connected set their approach requires. 

When the authors regress the estimated fixed effects on institution characteristics, the effect of university or college prestige is always small and generally insignificant. The authors find some evidence that institutions with larger endowments per student pay a modest premium. The authors repeat the exercise, replacing the two-step estimation with a single step that includes institution characteristics rather than institution fixed effects. The results are similar, as expected since both approaches provide consistent estimates of the same parameters. 

The authors also examine the correlation between institution prestige (measured by rank) and faculty quality, as measured by the individual fixed effect. The correlation is positive, consistent with our expectations (and probably those of most faculty at research universities). The authors briefly discuss how to reconcile the absence of a prestige premium, the positive match between prestige and faculty quality, and our sense that faculty at prestigious institutions would earn less at less prestigious institutions. The authors develop a toy hedonic model in which faculty transition only among similarly ranked institutions. 

The authors conclude with some thoughts about why our results differ from AKM models of the broad labor market.