EBITDA’s dominance over EBITA in explaining valuations has been declining over time, while the performance difference between EBITA and EBIT has been increasing. ssrn.com/abstract=29996…
"In terms of ability to predict stock returns, it appears that a structural change has occurred after the financial crisis, as the three operating income measures have failed to consistently predict stock returns over the last seven years."
"percentage pricing errors (%PE) corresponding to valuations using EV multiples of operating income performed very well in predicting stock returns over the 30 years sample period. Specifically, low %PEs (ie, low [price – estimated value] / price) predicted high stock returns..."
Robust: "The stock return predictive ability of EV multiples of operating income is monotonic across the %PE portfolios, and it holds after controlling for the five Fama and French (2015) factors as well as the momentum factor (Carhart, 1997)."
Although EBITA is not (yet) commonly used, this practice is quite prevalent
nowadays, especially in company presentations. A related practice is measuring the rate of operating profitability in terms of return on tangible capital instead of or in addition to ROIC:
nowadays, especially in company presentations. A related practice is measuring the rate of operating profitability in terms of return on tangible capital instead of or in addition to ROIC:
"the cash flow statement D&A includes amounts that were capitalized into the ending balance of inventories, which is reported on the balance sheet rather than being expensed. Therefore, adding to EBIT the full amount of D&A from the cash flow statement overstates EBITDA."
"when companies acquire existing businesses, the acquired intangibles are recognized as assets and, if they have finite lives, are amortized. This may lead to understated EBIT": 
Valuation multiples: "Research demonstrates that the harmonic mean performs significantly better than other location measures (e.g., mean, median), and that the mean performs particularly poorly (e.g., Liu, Nissim, and Thomas, 2002)."
Beware of capital changes: "Multiples of flow metrics—such as P/E or EV/EBITDA—compare price at the end of the period (P or EV) with a flow (EPS or EBITDA) that is generated during the period...if that capital changes during the period, the multiple is measured with error." 
As shown in Panel A of Table 3, there is a strong monotonic negative relationship between the portfolios’ %PEs (pricing error) and returns (PRet), indicating that valuations using EV/EBITDA multiples performed very well in predicting stock returns. 
"Interestingly, the %PE portfolio returns are strongly negatively correlated with the momentum factor, suggesting that EV/EBITDA multiple valuation enables one to differentiate between stocks for which return momentum is likely to continue from those likely to reverse."
