Assessing the Construct Validity of Alternative Proxies for Expected Cost of Equity Capital
Christine A. Botosan
University of Utah - School of Accounting and Information Systems
University of Utah - School of Accounting
Many research questions of interest to the accounting community cannot be adequately addressed in the absence a valid proxy for expected cost of equity capital. As "true" r is inherently unobservable, the ability of empirical research in this area to produce useful inferences depends on whether the proxies for r employed in the analysis measure what they purport to measure. Empirical evidence regarding the validity of various proxies employed in the literature is limited. Instead it is often assumed that these proxies reliably capture cross-sectional variation in r with little evidence provided to support this conjecture.
In this paper we assess the construct validity of four alternative proxies for r using a standard statistical approach to construct validation. The four proxies for r selected for analysis in this study are ones used in existing empirical research and are derived from well-accepted asset pricing methodologies. We first assess the extent to which the estimates produced by these four proxies are correlated with CAPM beta and firm size in the manner predicted by theory. We then examine the robustness of our results to the inclusion of expected earnings growth, a potential correlated omitted variable. Finally, we undertake three additional analyses: (1) examining the stability of the estimates produced (2) assessing the robustness of our primary results to the inclusion of additional firm characteristic variables and (3) examining the robustness of the relationships between the proxies and our risk factors for subsets of the sample. In general, we find that two of these proxies, one based on an internal rate of return derived from estimating the classic dividend discount model (DIV) and one based on the price-earnings-growth relationship (PEG), appear to capture cross-sectional variation in r. By providing evidence of the extent to which various proxies for r measure what they purport to measure, the analysis presented in this paper makes an important contribution to the accounting literature.
Number of Pages in PDF File: 42
Keywords: cost of capital, risk, terminal value, validity, dividend discount model
JEL Classification: M41, G12, G31
Date posted: May 21, 2002