1. Introduction
Firm value is shifting from tangibles to intangibles with a significant increase
in activities of research and development (R&D) (i.e., Lev and Zarowin, 1999) and
there has been a substantial increase in reported losses (Hayn, 1995; Joos and Plesko,
2005; Skinner and Soltes, 2009). These two variables are highly related. For example,
Joos and Plesko (2005) find that loss firms contain more R&D investments, and Li’s
(2011) untabulated results show that, among these loss firms, R&D are negatively
associated with forecast earnings. Their results raises questions for us: Why do
negative earnings, which imply a low marginal return to capital, encourage rather than
discourage R&D investments? Based on Hayn (1995), who proposes that loss itself
could persist for several years, I wonder whether loss firm’s R&D is an important
factor of loss persistency. That is, do loss firms invest in unproductive R&D? If so,
what factor drives them to do this and how to mitigate such relationship?
Prior research in organization literature has provided behavioral theory (Cyert
and March, 1963, henceforth BT) and prospect theory (Kahneman and Tversky, 1979,
henceforth PT) for us to address our research questions. However, in my viewpoint,
the above two theories are not yet integrated well to an intact theory to explain the
association between loss and R&D, and further, its implication to future earnings. To
illustrate, BT suggests that firms failing to attain a certain level of performance may
2
seek to solve the problem by investing in R&D (in their term, failure-induced R&D,
Chen and Miller, 2007; Chen, 2008), and their tolerance for risk may increase at the
same time; however, the increased risk tolerance influences firms’ behavior on
"high-level decision making" rather than "R&D decision"1. In other words, BT
attributes failure-induced R&D to the problem-solving incentive and excludes the
risk-seeking incentive (in my term, prospect incentive) from R&D decision process.
Respecting PT, except for giving formal proof on risk-seeking behavior when firms
are under a certain performance level (Kahneman and Tversky, 1979), Bowman (1980)
shows that such behavior may trigger undesirable results, i.e., a negative association
between risk and return for firms. However, the above findings of PT have not been
applied to explain the failure-induced R&D. In brief, although both BT and PT
address the problem of firm's decision making, their main interests are different. BT
examines firm's risky change, such as R&D, under different performance level and
excludes the prospect incentive from R&D decision; while PT, rather than examining
firm's R&D activities, it focuses on the relationship of prospect incentive (risk taking
behavior) under different performance level and its subsequent financial results. This
discrepancy restricts researchers under BT to predict the future performance of
failure-induced R&D because it does not attribute such R&D to prospect incentive but
1 To see the detailed explanation, please see section 2.2.
3
to problem-solving incentive, which lacks theoretical support upon the link of future
performance.
As such, I try to bridge this gap by incorporating PT into BT, and therefore
provide in-depth analyses on the following issues. (1) Since my investigation is on
loss-R&D relationship, I first adopt the failure-induced R&D framework suggested by
BT, and consider whether BT can be applied with a simple zero earnings point, which
serves as a reference level for each firm, rather than a complex mixture of both firm
itself and its competitors’ recent and past performance. (2) Whether prospect
incentive acts as an important factor in failure-induced R&D, i.e., BT overlooks the
importance of prospect incentive in R&D decision since it excludes prospect incentive
from R&D decision. (3) Whether a negative association between future earnings and
prospect-based R&D (i.e., the increased R&D which is based on prospect incentive)
exists since PT predicts that firm's prospect incentive (risk taking behavior) under its
reference point (aspiration level) deteriorates future earnings. (4) From the accounting
viewpoint, I wonder whether there exists mechanism to curb such undesirable and
inefficient activity, thereby providing a potential moderator, financial reporting
quality, and testing whether it mitigates firms’ (those with higher prospect incentive)
inclination to overinvest in R&D.
4
I therefore derive the hypotheses corresponding to each issue as follows. First,
I expect BT can be applied with a simple zero earnings point, which serves as an
reference point for management, because accounting literature has confirmed its (zero
earnings point) importance by showing that management inclination to avoid losses
(Graham et al., 2005; Hansen, 2010; Baber et al., 1991; Burgstahler and Dichev, 1997;
Gunny, 2010). Second, I expect that prospect incentive drives firms with big loss to
invest more in R&D compared with those in small losses because R&D is an
investment with high uncertainty (Kothari et al., 2002) and firms tend to adopt
projects with high risk when their losses become larger according to PT. Third, based
on prior research, showing that risk taking behavior under adversity incurs worse
future earnings (Bowman, 1980; Fiegenbaum and Thomas, 1988; Chang and Thomas,
1989; Fiegenbaum, 1990; Jegers, 1991; Sinha, 1994; Gooding et al., 1996; Lehner,
2000), I expect that the prospect-based R&D deteriorates future earnings. Forth,
accounting literature has shown that both losses and R&D activities will increase the
possibility of moral hazard, which in turn deteriorates future accounting performance2.
To illustrate, once firms are in losses, they have great incentive to meet/beat zero
earnings benchmark through manipulating discretionary accruals (e.g., Hansen, 2010)
and real activities (Baber et al., 1991; Burgstahler and Dichev, 1997; Gunny, 2010);
2 To see the detailed demonstration, please see section 2.4.
5
while investments in R&D will exacerbate the degree of information asymmetry
between firms and their shareholders/debtholders (Clinch, 1991; Myers and Majluf,
1984; Aboody and Lev, 2000), which in turn increase the likelihood of moral hazard
in which firms engage (Arrow, 1971; Pauly, 1974; Holmstrom, 1979; Healy and
Palepu, 2001). Moral hazard may thus lead loss firms to engage in unproductive
activities by driving them to overinvest R&D since their inclination to take undue risk
increases. Since high quality financial reporting mitigates moral hazard by facilitating
contracting and monitoring (Kanodia and Lee, 1998; Healy and Palepu, 2001), I
therefore expect that reporting quality curbs unproductive R&D.
Distinct from prior studies, which employs the absolute difference between firm’s ROE and its industry’s contemporaneous median ROE as the measure of firm’s
risk taking behavior (Miller and Bromiley, 1990; Fiegenbaum, 1990; Jegers, 1991;
Sinha, 1994; Gooding et al., 1996), I create a simpler proxy based on the fourfold risk
attitudes proposed by PT to capture firm’s risk taking incentive (prospect incentive).
Measures of other variables are largely consistent with literature.
Based on a sample of 19,788 firm-years from 1993 to 2006, I find that loss
firms increase their R&D spending, implying that zero earnings point serves as an
aspiration level for firms (H1). Further, I find that firms with big loss invest more in
R&D than those in small losses, indicating that prospect incentive (based on fourfold
6
risk attitudes framework) drives loss firms to invest more on R&D (H2). Respecting
the future performance of prospect-based R&D, I find that prospect-based R&D of
loss firms deteriorates future earnings (H3). Further analyses show the effect of
prospect incentive on R&D is less positive for firms with higher quality financial
reporting, providing support for H4. These results hold after addressing the potential
sample selection bias of missing value of R&D excluding those observations instead
of resetting them to zero, and they are robust to using total assets and market
capitalization of common stock to scale R&D (operating income before depreciation,
advertising and R&D expenditures) as an alternative proxy for R&D (earnings). My
evidence highlights the important role of prospect incentive in loss-induced R&D and
thereby deteriorating firms' future earnings.
My study contributes to the literature along the following dimensions. First,
with respect to my research design, I capture firms' prospect incentive by employing
the characteristics of PT’s fourfold risk attitude argument which is different from the
traditional industry median ROE approach, and further, I incorporate it into BT to
explain the loss-induced R&D. By doing this, I show that prospect incentive exists
and does affect loss firm’s R&D decision significantly, indicating that extant research
may put overemphasis on problem-solving incentive, which lacks theoretical support
and is immeasurable, and overlook the importance of prospect incentive when
7
analyzing loss-induced R&D. Introducing prospect incentive into firm’s investment
decision extends the flexibility of the application of BT, i.e., researchers can examine
the investment projects either with high risk or with low risk. To illustrate, since firms
at different performance level may differ in risk preference, researchers can use this
characteristic to predict firms' choice of investment projects, which cannot be
achieved by existing BT framework. I also elaborate prior studies by presenting more
detailed analyses on the R&D-aspiration level relationship with a PT perspective. In
other words, I not only examine the effect of performance below/above the aspiration
level on R&D but also examine whether firm’s R&D investments vary among each
region. Most importantly, my approach provides a framework to understand the
implications of loss-induced R&D on firm’s future accounting performance. To
illustrate, since I emphasize the importance of prospect incentive on loss-induced
R&D, rather than problem-solving incentive, I have a strong theoretical background
(Bowman, 1980; Fiegenbaum and Thomas, 1988; Chang and Thomas, 1989;
Fiegenbaum, 1990; Jegers, 1991; Sinha, 1994; Gooding et al., 1996; Lehner, 2000)3
to predict there is a negative association between loss-induced R&D and future
3 Researchers have found that firms in adversity and with excessive risk taking behavior perform
worse in future. The possible reasons proposed by Fiegenbaum and Thomas (1988) are failed firms' inclination to escalating commitment, less diversified strategy, angency problem and organization inertia.
8
earnings, which cannot be achieved by existing BT framework. Hence, my study
develop a new framework that can explain firms with different risk attitude invest
differently in R&D and therefore lead to different financial results all at once, again,
which cannot be achieved by existing BT framework.
Second, I find a new variable, prospect incentive4, moderating R&D’s future
financial performance. Specifically, I show that as firm's losses become larger, which
is accompanied by raising prospect incentive, R&D activities reduce the positive
association between R&D activities and subsequent accounting performance. That is,
while investments of loss firms can be those firms’ attempt to revert to normal, as
suggested by the abandonment option of Hayn (1995), their R&D expenditures fail to
help them upon such objective. My results supplement Hayn (1995) by showing the
actual performance of loss firms' investments and this more negative association
between loss firms’ R&D activities and subsequent accounting performance is
consistent with Li (2011), who suggests that R&D activities of loss firms is negatively
associated with forecast earnings. My results also extend prior studies (e.g., Lev and
Sougiannis, 1996) that show a positive association between all firms’ R&D
4 I use the interaction of two dummy variables, LOSS and BIG, to capture the prospect incentive since firms' prospect incentive raises as their losses become larger, as per PT. The former is a dummy variable equal to one, if the earnings is negative, and the latter represents that firm's earnings to one-year lagged total assets is above (below) the industry's median value of profit (loss) firms’ earnings to one-year lagged total assets in year t.
9
expenditures and subsequent earnings. It therefore may help investors improve their
prediction on firms’ future performance.
Third, my study adds to a growing empirical literature in accounting that
examines how financial reporting quality affects investments (e.g., Bushman et al.,
2005; Biddle and Hillary, 2006; McNichols and Stubben, 2008; Biddle et al., 2009;
Chen et al., 2011). Distinct from prior studies, which examine the effect of reporting
quality on the efficiency of “total investment” or “capital investment” (e.g., Biddle
and Hillary, 2006; Biddle et al., 2009), my study examines the effect of reporting
quality on the efficiency of “R&D”. To illustrate, my evidence shows that higher
reporting quality alleviate firms’ (those with higher prospect incentive) inclination to
overinvest in R&D, which in turn lead to worse future financial results as Ming-Liang
et al. (2010) suggested.
The rest of the thesis is organized as follows: Section 2 develops the research
questions. The research design is presented in Section 3. Section 4 presents the
sample characteristics. Section 5 presents results, and Section 6 concludes the paper.