“A company's market value is a function of its book value of equity, earnings, and other information” (Ohlson, 1995). One can either interpret firm value as the risk premium explained by common market risk factors in asset pricing theory (Fama & French,
1996–2008; Ali, Hwang, & Trombley, 2003; Baginski & Wahlen, 2003). Alternately, firm value can also refer to operating performance based on summation of accounting numbers in fundamental analysis (Ohlson, 1995; Myers, 1999). Although performance sometimes fails to be realized as market stock price, the reason business valuation from market and operating perspectives is too distinct remains unexplained.
No one knows the intrinsic value of firms under efficient market hypothesis (EMH).
Scholars believe market stock price to be the best reflection of intrinsic value. We can thus evaluate the degree of asymmetric information through the explanatory power of valuation indicators to market stock price in relative magnitude of adjusted R-square or correlation coefficients (Yoo et al., 2004; 2008). Most studies often focus on one question: “Which valuation indicator is more informative due to lower asymmetric information?” In the present paper, we employ three market stock returns (Ri, ER1, and ER2) as proxies for performance from the market perspective (Fama & French, 1998–2008; Bhagat & Bolton, 2008). In addition, we use two earning-based measures, EPS and EVA1
The pros and cons of EPS have been well documented in many studies and textbooks (Ross, 2006). EPS is considered as the shareholders’ wealth, and more reasonably related to
, as proxies for valuations from the operating perspective to illustrate the information content between EPS and EVA.
1 EVA (economic value added) is also known as residual earning or residual income (RE or RI). Basic formula of EVA in per share basis (1) is listed below:
EVA t = EPS t － r × BPS t-1 , where EPS t denotes forecasted EPS at time t-1; r denotes Implied (opportunity) Cost Of Equity (ICOE)2; BPSt-1 denotes Book value of equity funds per share at time t.
the goal of maximizing firm value. Thus, EPS is considered more highly associated with both returns and firm values than EVA or cash flow from operations (Biddle, Bowen,
&Wallace, 1997). In addition, because the computation of EVA contains EPS, the
shortcoming of EPS is passed on to EVA as well (Ohlson, 2000). Major problems of EVA can be demonstrated by its basic formula in per share basis in notation 1.
The first problem is forecast EPS. Although forecast EPS is not available at all times in all firms, some researchers prefer to articulate the role of forward EPS in valuation. They state that forward EPS is more informative than EPS, EVA, DCF, and DDM (the worst is listed last) (Ohlson & Juettner-Nauroth, 2000). Meanwhile, according to Richardson and Tinaikar (2004), there exist detective links between historical and forecast data branches, which often produce similar results. Moreover, long-term analyst earning forecasts into EVA have been proven not to improve pricing performance significantly (Lo & Lys, 2001).
Some, such as Frankel and Lee (1999), continue to use shorter forecast horizon with one- and two-year ahead analyst earnings forecasts; however, these still suffer from biases in forecasting errors. The limitations encountered by previous studies suggest the validity of using historical EPS over forecast EPS in this paper. The second problem with EVA concerns ICOE2. ICOE must be estimated, and is often viewed exogenous. In the present study, we use CAPM-derived ICOE because individual betas predict positive and
systematic association to ICOE in the literature (Yoo et al., 2004)3
2 ICOE here is derived from CAPM, which substitutes beta for individual firms (βi) in models because individual betas predict positive and systematic association to ICOE in literature. Several alternative approaches to estimate ICOE are well reviewed in Section 2.1 of Yoo et al. (2004).
. Likewise, we obtain CAPM-derived ICOE by substituting beta for individual firms (βi) into CAPM. The third problem with EVA involves clean surplus relation (CSR) violation. According to Ohlson
3 Several alternative approaches to estimate ICOE are well reviewed in Section 2.1 of Yoo et al. (2004): (1) ex post realized stock returns as natural proxy for the ex ante ICOE, but proven noisy and potentially biased (Fama & French, 1997; Elton, 1999); (and 2) internal rate of return (IRR) that equates stock prices with the valuations based on analyst earnings forecasts (Gebhardt et al., 2001).
Most variation of stock returns can be described by FF3 or FF4 (above 80% in US market), however, stock returns prove to be noisy; thus, we neglect FF4-derived ICOE.
(2000), CSR violation could affect EVA on a case-by-case basis.
Despite problems of EVA above, some studies continue to align with EVA to be more informative. Frankel and Lee (1999) conclude that firm value estimates derived from EVA can better explain the cross-sectional distribution of the stock prices, accounting for more than 70% of its variation within 20 countries, than earnings or book value. Even for studies that claim EPS outperforms EVA, empirical results continue to reveal that EVA has
significant marginal contribution (Biddle, Bowen, & Wallace, 1997; Lundholm, 2001; Yoo et al., 2004; 2008)4. Therefore, EVA has information unique from EPS. In addition, the authors suggest that EVA and AEG (abnormal earning growth) models are more related to systematic and industrial impact and risks (Jeon, Kang, & Lee, 2003, 2005; Cheng, 2004;
Baginski & Wahlen, 2003). Yoo et al. (2004; 2008) indicate that these risks possibly represent economic-wide and country-level, industry-specific5
In sum, early studies all show that EPS is more informative (Biddle, Bowen, &
Wallace, 1997; Ohlson & Juettner-Nauroth, 2000). However, a phenomenon of conflict results on information content is evident approximately after year 2000. Recent studies show that EVA is more informative otherwise (Frankel & Lee, 1999; Yoo et al., 2004, 2008). The cause of such changes in the results remains unknown.
related information. Some suggest EVA may be more sensitive to ICOE because of its inner beta estimation in the hotel business; however, there is a lack of significant empirical results. We argue that EVA can reflect some information from the market.
Because of the coincidence of time with conflict in studies involving EVA, we consider whether this can be caused by the rising angle of behavior finance. Behavior finance provides a fresh viewpoint to review many issues, especially in light of Kahneman’s
4 Biddle, Bowen, and Wallace (1997), using incremental tests, suggest that EVA components can add
marginally to information content beyond earnings. Lundholm (2001) suggests that, under certain assumptions, DCF and EVA can even carry similar information. Yoo et al. (2004, 2008) list facts regarding uncertain dominance of EVA depending on clean surplus relation (CSR) in global case.
5 Gode and Mohanram (2003) and Guay, Kothari, and Shu (2003) claim ICOEs derived from EVA (RIV models in original paper) reflect more “industry-specific” information.
Nobel Prize in 2002. Kahneman’s study is more committed to ideas regarding psychology such as anticipation, bias of analysts, and mental account (Tversky & Kahneman, 1981;
Thaler, 1985). The studies on behavior finance prove to against the impact of EMH-based studies. In the present study, we aim to determine whether behavioral finance effect also affects our valuation.
To be more specific, in business valuation, EMH and behavioral finance have different point of views regarding market reaction. EMH believes business operating information is fully reflected in stock returns, whereas behavioral finance state stock returns have
leadership in business information over EMH. Therefore, some studies suggest behavioral finance effect is likely to have power for reversing the directions of causality between stock returns and earnings (Bar-Yosef, Callen, & Livnat, 1987; Peiers, 1997; Linnainmaa, 2010).
This can be realized by dot-com companies during Internet bubbles. For companies such as America Online and Twitter, which have encountered long-time losses but still receive high market prices6
As a result, this study mainly develops to answer three questions: First, as previous studies did, we want to verify which earning indicator makes asymmetric information decrease the most. Further, we want to investigate whether the blooming of behavior finance does make a changing viewpoint of century, and whether we can directly see the behavior finance effect in time and in causality. Finally, can we then document which
, or for corporate companies such as Netscape and Facebook, which have new business models and lack of accordance of past performance, the earning-based valuations seem to be off the hook to the stock returns; otherwise, even though valuations are affected by stock returns, they do not correspond to the positive relationship between earnings and returns that we expect under EMH.
6 In the Internet bubble in 1998 to 2000, as Penman (2010) states, all dot-com companies worth over 1 trillion dollars in total, with 33 times on average price-to sales ratio far beyond the historical level of 1, maintain only 30 billion dollars on revenue; a recent report by The Guardian UK (2011)) also suggests a 2nd phenomenon of Internet bubble may occur. Microblogger and Twitter have an estimated worth of $ 10 B; Facebook, which have recently planned to initiate IPO, is estimated to be worth $ 60 B; this figure is slightly over Ford ($ 55 B) and below Visa ($ 63 B).
earning indicator is more influenced by behavior finance effect, and thus affecting its information content.
Several firm characteristics, including financial constraints, growth of investments, profitability, solvency, and debt ratio, are viewed as control variables. These variables are not Fama and French factors, but are also recognized as existing influences on firm value in extensive prior studies7
The remainder of this paper is organized as follows. Section 2 describes the methodology, including the sample selection, research models, and beta estimations.
Section 3 presents and discusses the empirical results. Section 4 provides a summary of our main findings and the conclusion.
. Some studies show that the information regarding these influences may not be possibly explained by FF factors. We specifically intend to sort out the impact of financial constraints (proxy by asset tangibility) on firm value, thereby showing that investing behavior changes under financial constraints.