Figure 4 summarizes our overall findings on the value-relevance of EVA versus the
other performance measures. It combines comparisons of relative information content
comparisons (Adj. R2s) from question 1 (which are shown as circle sizes in the figure), and
comparisons of incremental information content comparisons (F-statistics) from question 2,
shown as relative positioning or lack of overlap (with less overlap indicating more incrementalinformation content). Overall, neither EVA nor RI appears to dominate NI in its association
with stock market returns. NI has the largest relative information content (as indicated by the
largest circle) and the overlap between circles is large suggesting that there is little incremental
information content in EVA, RI and CFO beyond that contained in NI.
[Insert Figure 4 here]
We also examine a related claim that EVA is more highly associated with firm values (versus
stock returns considered above):
Q3: Does EVA dominate earnings in explaining firm values?
To address this question, we replicate and extend a study authored by former Stern Stewart vicepresident
Stephen O’Byrne that appeared in the Spring 1996 issue of this journal.17 O'Byrne first
compares Adj. R2s from regressing firm value on EVA and earnings measured as NOPAT. He
reports an Adj. R2 of 31% for the EVA regression and 33% for the NOPAT regression. Next, he
adjusts the EVA regression by: 1) allowing separate coefficients for positive and negative values
of EVA, 2) including the natural log of capital in an attempt to capture differences in the way the
market values firms of different sizes, and 3) including 57 industry dummy variables in order to
capture potential industry effects. When all of these adjustments are included, O’Byrne obtains a
larger Adj. R2 for the enhanced EVA regression (56%) than for NOPAT (33%).
However, notice that O'Byrne makes ‘adjustments’ only to the EVA regression. When we
‘level the playing field’ by applying the same adjustments to the NOPAT regression and also
examine NI, EVA’s superiority disappears. The NI regression has a significantly higher
association with firm value (Adj. R2 = 53%) than the EVA regression (Adj. R2 = 50%) and the
NOPAT regression (Adj. R2 = 49%). The latter two are statistically indistinguishable. Thus, aswith our stock returns tests, these results do not support the contention that EVA outperforms
earnings in explaining firm values. To the contrary, and in contrast to claims by Stern Stewart,
our evidence suggests that earnings more often dominates EVA in value-relevance to market
participants.