“…This lesson is supported by Rasmussen and Zuehlke's rejection of a BoxCox in favor of a quadratic specification. Moreover, Yinger (2010) shows that the Box-Cox, log, and semi-log specifications are all inconsistent with sorting, at least given constant-elasticity demand functions for S and H. Overall, the inconclusive and sometimes contradictory results in these studies remind us that more research is still needed to guide the choice of functional forms in empirical studies of school quality capitalization. At this point, it seems reasonable to recommend that capitalization studies should at least examine a specification including S and S 2 , which is consistent with Epple's (1987) and with Yinger's third special case in Eq.…”