Measurement Errors of Expected Returns Proxies and the Implied Cost of Capital

Abstract:     
I develop a novel diagnostic procedure to estimate the associations between measurement errors of expected returns proxies and firm characteristics. Application to GLS, a popular implementation of the implied cost of capital ("ICC"), yields the first direct empirical evidence that ICC measurement errors i) are persistent, ii) can be associated with firms' risk or growth characteristics, and therefore iii) can lead to spurious inferences in regressions. I devise a novel methodology to account for the influence of ICCs measurement errors in regression settings, and show that its application i) can explain some puzzling associations between GLS and firm characteristics and ii) can improve upon GLS, by forming new ICCs that better sort realized returns. Together, the innovations of this paper allow researchers to better understand ICC measurement errors and provide a robust empirical strategy for future research.

Number of Pages in PDF File: 63

Keywords: Expected returns, implied cost of capital, measurement errors

JEL Classification: O16,G30, D03, P34

working papers series